Objective: There has been at least 40 years of active research on screening for depression and distress in primary care. Both successes and failures have been documented. The purpose of this focussed narrative review was to summarise this research and present the key lessons for clinicians and researchers working in psychosocial oncology.
Methods: We searched for studies assessing the utility of screening in primary care in seven electronic bibliographic databases (CENTRAL, CINAHL, Embase, HMIC, Medline, PsycINFO, Web of Knowledge) from inception to December 2010. Results were reviewed and summarised into key areas.
Results: We found that research could be distilled into the following key learning points. (1) Primary care is an important partner in psychosocial care. (2) Both over and under detection are problematic. (3) Barriers to identification involve patient and clinician factors. (4) Acceptability of screening is critical to implementation. (5) Underserved groups need special attention in screening. (6) Patient–clinician trust is an important modifiable variable. (7) Greater contact influences detection. (8) Clinician confidence/skills influence screening success and subsequent action. (9) Training may improve confidence but effects upon long-term outcomes are modest. (10) Screening is generally ineffective without aftercare.
Conclusions: Primary care has shown largely what does not work in relation to screening. Namely relying on clinicians' unassisted judgement without infrastructural support, using over-complex scales with low acceptability, looking for depression alone, using screening without linked treatment, treating in the absence of follow-up and failing to engage patients in their own care. These pitfalls can and should be avoided in psychosocial oncology. Copyright © 2011 John Wiley & Sons, Ltd.
Primary care has long been known to be the de facto mental health system and thus has been a key venue for the testing of depression scales and implementing screening. Several early randomised and non-randomised studies in primary care tested the effect of provider receipt of screening information (versus not receiving screening information) 1, 2. In general, these studies were disappointing, as primary-care providers either failed to make use of screening results to initiate treatment or referral for depression or, if they did, there were no differences in symptom levels 3. Despite these failed studies, the advent of the selective serotonin reuptake inhibitors in the 1990s brought on a new era for depression treatment. These medications were much easier to prescribe, highly accessible and reasonably well tolerated. Additionally, the makers of these medications embarked on major public education campaigns, which in combination with governmental campaigns (e.g. Depression Awareness, Recognition and Treatment initiative in the United States, and the Defeat Depression campaign in the United Kingdom), increased public awareness of depression as a treatable medical illness. This renewed interest in depression treatment led to the development of a number of assessment tools developed specifically for use in medical settings (e.g. PRIME-MD, PHQ9, SDDS-PC and BDI-SF). In this second phase of screening (1990–2009) identification was combined with organised treatment and follow-up in an attempt to influence quality of care. Although there were benefits, there were also limitations to the approach that focused on depression alone. Recently colleagues in primary care have recognised the need to identify multiple mental health conditions including but not limited to depression. We are now beginning a new phase of screening in which outcomes such as distress, unmet need, desire for help and functional limitations are key. These are likely to be supported in DSMV and ICD11, which may place more emphasis on distress and dysfunction across mental disorders.
Against this historical backdrop, there have been a number of valuable lessons learned concerning what works and what does not work in primary care. With the growing realisation that depression and medical illness are intertwined, and that depression can negatively influence medical illness, it is timely to take the lessons learned from primary care and apply them to other medical settings. Nowhere is this more appropriate than in cancer settings, where patients suffer high rates of distress, endure complex and burdensome treatments, and live with an uncertain but feared future. Our aim therefore was to perform a narrative review of studies conducted in primary care pertaining to the clinical utility of screening and detection of depression and distress and extract the most useful lessons for cancer care.
Data sources and searches
We searched for studies assessing the utility of screening in primary care in seven electronic bibliographic databases (CENTRAL, CINAHL, Embase, HMIC, Medline, PsycINFO, Web of Knowledge) from inception to December 2010. The search was kept as broad as possible with search terms ‘screening, or case-finding or identification or diagnosis’ and ‘depression or distress or mental or psych*’. We filtered hits for primary data studies. Additional papers were found by searching the references of retrieved articles, tables of contents of relevant journals, previous systematic reviews and meta-analyses of screening and case identification for depression. We did not extract studies on issues beyond depression/distress, on delivery of psycho-oncology services in primary care or primary-care-based interventions in general which were all considered beyond the scope of this review.
Data extraction and quality assessment
Data were extracted by two researchers and summary study information characteristics extracted were country of study, setting, patient characteristics (e.g. age and gender), scales used to identify depression, reference standard, and most importantly the key messages for cancer clinicians. We sorted findings into 10 themes, or key lessons.
Lesson 1. Primary care is an important partner in psychosocial care
Integration of primary care and specialist health care is increasingly seen as important, particularly for continuity of care 4. Each year an estimated 80% of the population consults a general practitioner (GP), and between 20 and 40% of these visits involve a mental health concern 5. About half of these can be considered new episodes 6. About two-thirds of all psychosocial appointments and two-thirds of antidepressant prescriptions are generated in primary care 7, 8. Importantly primary-care services are often perceived as more accessible, less stigmatising and generally more acceptable to patients 9. This appears to be consistent in different regions even with wide variations in health-care delivery.
As with oncology settings, depression and anxiety are the predominant common mental health problems seen by GPs 10. That said, most people presenting do not have depression or anxiety and thus these disorders can be hard to detect 11–13. Typical prevalence estimates are about 15% for major depression with 10% who have minor depression, 25% overall 14. Further most depression in primary care is of mild-to-moderate severity 15. Detection of less severe disorders is a particular challenge because symptoms do not differ greatly from healthy but stressed individuals 16, 17. A similar distribution is seen in cancer settings. In the Leicester Cancer Centre we found that the rates of mild, moderate and severe depression on the HADS-D were 11.6, 7.3 and 4.0% among 688 chemotherapy attendees 18. In total, 50.8% of those with symptoms of depression were considered ‘mild’ in severity. In primary care Thompson et al. examined the relationship between severity of depression on the HADS-D and proportion of cases detected 15. Although a higher severity of depression was associated with greater recognition, 50% of all correct identifications occurred at a HADS-D score of between 8 and 10 because of the greater proportion of mild depression. Just like specialist settings, in primary care at least 75% of individuals with depression have other comorbidities 19–21. Yet despite this partnership GPs struggle to deal with mental health problems and often offer inadequate treatment or follow-up 22, 23.
We suggest these issues may also be present in cancer settings and that it is appropriate for multidisciplinary clinicians working in these settings to embrace the management of depression in their patients. Because cancer clinicians often know their patients well and attempt to improve trust they are well suited to take responsibility for the initial treatment of depression in the context of cancer care 24. Additionally, patients often feel more comfortable staying within this setting where a high degree of trust is likely to have developed.
Lesson 2. Both Over and Under Detection are Problematic
Convincing data show that clinicians in all medical specialties have difficulty recognising mental disorders 25, 26. This is the case even in diverse health-care settings 27. Recently a large-scale meta-analysis summarised the rates of correct identification in primary care 28. A typical GP in an urban practice where the prevalence is 20% would correctly identify 10 out of 20 cases, missing 10 true positives. Tiemens et al. (35) found that only 26% of these missed cases were complete omissions, 25% were underestimates of severity (e.g. diagnosing subthreshold instead of mild) and 38% were misidentifications. In the MAGPIE study, Bushnell 29 found reasons for false negatives were not considering psychological issues as clinically significant (23.4%), recognising clinical significance but not ascribing a diagnosis (7.1%) and misidentification (7.7%). Under recognition may be converted into under treatment if such individuals are not re-examined at a later date because recognised patients are more likely to be offered mental health interventions 30. Further if screen positive patients are not offered help, about two-thirds will have significant depression on long-term follow-up 31. Data from the European Study of the Epidemiology of Mental Disorders showed that only 15.1% of those with an identified mood disorder and 23.2% with an anxiety disorder received either drug or psychological treatment 12. Similarly, Maginn et al. found that GPs recorded active management of a psychological problem in 37% of patients whom they rated as cases 32. Of these, 24% were prescribed psychoactive drug treatment, 5% were referred to psychiatric or psychological services and 3% were offered both drug and psychological treatments. Under treatment of unmet needs has yet to be thoroughly studied in cancer care, but one study showed that only 12% of women meeting criteria for major depression reported currently receiving medications for depression, and only 5% of women reported seeing a counselor or participating in a cancer support group 33. In a second study from the large US National Health Interview Survey (n = 4878) only 34.6% of cancer patients with mental health needs had received services in the last year 34.
Just as important as missed cases are false positives. False alarms can lead to inappropriate treatment if there is inadequate follow-up of suspected cases. Mitchell et al. showed that clinicians attempting to spot depression would actually make more false-positive errors than false negatives 28. In a busy urban practice, he/she would be likely to correctly re-assure only 65 out of 80 non-cases, falsely diagnosing as many as 15 people as depressed. Overall the fraction correct would be 75% with a misclassification rate of 25%. From the clinicians' perspective out of every five cases thought to be depressed, only two would be true cases but 6 out of 7 non-cases would be correctly re-assured. Reasons for false positives have been studied in primary care. Approximately 35% are typically overestimates of severity, 24% misdiagnoses and 41% complete errors 35. It is likely that the same issues with false positives are represented in cancer settings 36. The key lesson then is that all clinicians must be aware of both types of error and further be aware of the limits of clinical accuracy.
Many cancers as well as their treatments produce a number of symptoms that contribute to distress and in some circumstance may mimic symptoms of depression contributing to either over or under recognition 37. If educated about symptoms of depression, cancer specialists are well positioned to look for these symptoms and thereby improve the accurate assessment of depression in their patients.
Lesson 3. Barriers to identification involve patient and clinician factors
In primary care there have been several impressive studies examining predictors of correct detection and these provide valuable lessons for cancer clinicians (Box 1). Borowsky et al. studied 19 309 patients from 349 primary-care physicians using the MOS 8-item Burnam screen for depression and Diagnostic Interview Schedule for DSMIII 38. Physicians were less likely to detect depression in African Americans, men and those younger than 35 years and more likely when comorbid hypertension or diabetes was present. Hickie et al. looked at 46 515 patients attending 386 GPs of whom 56% were not recognised 39. This is probably the most comprehensive study of predictors of recognition available. Patients were more likely to be assessed psychologically if they were middle-aged, female, Australian-born, unemployed, single, presenting with mainly psychological symptoms or for psychological reasons. Doctor characteristics associated with willingness to assess were being aged over 35 years, having an interest in mental health, having had previous mental health training, being in part-time practice, seeing fewer than 100 patients per week, and working in regional centres. Thompson et al. (15) examined recognition among 156 GPs in the UK, involving 18 414 individuals using the HADS-depression subscale. Women, unemployed and those with anxiety were more likely to be detected while the elderly and retired more likely to be missed. In a nationwide study of 20 421 patients in Germany attending 633 primary-care doctors, Wittchen et al. found that recognition was associated with prior treatment episodes, increasing number of depression symptoms, patients' higher age, practice experience of greater than 5 years and the presence of psychomotor retardation 40. Pfaff and Almeida 41 found that 39.9% (87/218) were correctly classified as depressed by their GP. Older patients were more likely to be incorrectly classified as ‘not depressed’ by their GP when they were born outside of Australia or New Zealand, did not smoke or use sleeping tablets, acknowledged milder levels of depression and presented to their GP with primarily somatic complaints. Nuyen et al. found that among 191 depressed primary-care patients diagnosed using the CIDI, 28.8% were recognised and recorded by GPs over the same period 42. Patients without chronic somatic co-morbidity, a lower educational level, less severe depression, and fewer GP contacts all significantly increased the likelihood of not being diagnosed as depressed. Mccall et al. looked at predictors of recognition of distress in Australian primary-care practice 43. Twenty-eight GPs completed a clinical audit on 868 of their patients who completed the GHQ28. Correct recognition was associated with years experience as a GP, older age of patient and greater severity of distress. Several of these barriers to recognition have been replicated in cancer settings, but more research about predictors of good quality psychosocial care is needed 44–46. In both primary and secondary care more research is needed on predictors of false-positive diagnoses. In one study over-diagnosis was associated with higher levels of symptoms (SDS score), lower Global Assessment of Functioning scores, a previous history of depression and the absence of generalised anxiety 47.
The key lesson is that health-care organisations must recognise the complexities of detecting and managing mental ill health and attempt to reduce the barrier to good quality care. In fact, there are a number of efforts that reflect international agreement to integrate screening into routine cancer care 48–50. The same barriers that exist in primary care are likely to be present in oncology settings 44–46; therefore, it might be prudent to implement routine screening and assessment programmes to aid busy cancer clinicians who may otherwise overlook emotional distress.
Lesson 4. Acceptability of screening is critical to implementation
Only a handful of studies have looked carefully at the acceptability of screening and usually through the proxy of the willingness of clinicians to screen. Ming et al. observed 33 physicians and showed that physicians assessed depression in only 14% of visits and almost never used formal depression assessment tools 51. Bermejo et al. looked at attitudes to screening in general practice in South Baden, Germany using the Patient Health Questionnaire (PHQ9) 52. Patients rated the usefulness of the instrument more positively than GPs. Most GPs (62.5%) felt that the questionnaire was too long and 37.5% too time-consuming, even though it typically took only 1–2 min. Half (50%) of the GPs rated the PHQ as an impediment to daily practice and 75% thought it was impractical compared with only 25% of patients. Buist et al. surveyed almost 2000 health professionals involved in perinatal care 53. When broken down by profession, 90% of maternity nurses, 69% of midwives and 54% of GPs screened for perinatal depression in their routine practice, and across all groups 98% of these health professionals supported ongoing screening. From the patients' perspective discomfort with screening was higher in those with more severe depression, and 64% of those with elevated EPDS scores were uncomfortable with the screening process. Importantly, not all women agreed with their screening results with 18% disagreeing that they were depressed. Of note, in trials of screening programmes up to 50% of patients in primary care decline to participate in clinic screening interviews conducted by researchers or clinic nurses 54, 55. In general, short questionnaires have higher acceptability than longer ones 56.
In many ways psycho-oncology is leading the way in acceptability issues with innovations such as the distress thermometer and other short tools 57. The key lesson for all health-care settings is that no matter what the accuracy of a screening tool, it is often acceptability that determines successful implementation. Screening must remain acceptable to front-line staff if front-line staff are to be engaged in screening process.
Lesson 5. Underserved groups need special attention in screening
Culture, language, age and gender are common demographic factors that influence receipt of good quality care and the success of screening 58. Some ethnic groups appear more likely to communicate with clinicians about depression than others 59. It is known that there are major differences in psychopathology in different racial and cultural groups 60. Diagnosis may be more difficult in groups that tend to emphasise the somatic symptoms of depression over cognitive symptoms compared to white patients (e.g. African American and Hispanic American patients) 61, 62. Greater medical co-morbidity and higher somatisation may deter early diagnoses of mental health symptoms by primary-care clinicians or lead to misdiagnosis 63. Linguistic barriers to detection of depression may be significant and this can lead to inaccurate psychiatric diagnosis and greater misunderstanding between patient and clinician 64, 65. Maginn et al. examined GP recognition of distress in South London. Overall GPs identified 65%; however, Black African patients were less likely to be detected or treated than Black Caribbean and White English patients 32. Willingness to talk to the GP about psychological problems was the main predictor of detection. Ethnicity did not independently predict detection, but Black African individuals were less likely to talk to their GP about psychological problems. Even more worrisome, half as many Black African individuals with detected distress were offered treatment compared with English cases (41 versus 22%). Borowsky et al. conducted a study of 19 309 patients looking for factors that may increase the risk of non-detection of mental health problems in primary care. Despite meeting screening criteria for depression, both African American and Hispanic Americans were significantly less likely, compared to white patients, to have their depression detected by their physician (p = <0.01 for African American and p = <0.05 for Hispanic American) 38. Mauksch et al. conducted a study into low income and uninsured populations presenting in primary care. They found the most common health problem patients wanted to discuss with the care provider was a problem with mood (29%); however, only a third of these concerns were elicited by providers 66.
Poverty and low insurance are thought to be both risk factors for and consequences of poor mental health 67. Unfortunately uninsured, ethnic minority and low income populations may forgo routine medical check-ups and not attend or delay seeking medical attention due to cost concerns 68. Similar patterns are emerging in cancer settings. Ell et al. carried out a study on prevalence of depression in low income women with cancer. Women who were depressed had experienced more barriers to receiving cancer care. Barriers included financial inability to obtain all the medications, lost wages as a result of sick time, personal and family problems preventing keeping of appointments and also medical co-morbidity. These barriers to cancer care may also be barriers to seeking adequate mental health care in symptomatic but undiagnosed depressed patients 69. In summary, underserved and disadvantaged patients are apt to suffer the same problems regarding detection of mental health problems regardless of the clinical setting in which they are seen. Thus, it is incumbent upon providers (both oncologists and primary-care providers) to increase their vigilance and organise their practices so as not to overlook the mental health problems of these vulnerable patients.
Lesson 6. Patient–clinician trust is an important modifiable variable
According to patients, the most important needs are a clinician's interpersonal skills, the ability to recognise depression, the effectiveness of offered treatment and ability to reduce problems associated with treatment 70.
It should be of no surprise that recognition of distress and depression are linked with the number of symptoms disclosed during a consultation 32, 71. Recognition is facilitated when patients report psychological symptoms of anxiety or depression early in the consultation 72. Patients who normalise or minimise their symptoms are less likely to be identified 73. However, patients often do not complain of ‘depression’ and patients' views about their depressive symptoms are significantly different from conventional medical views 74, 75. Many authors have shown patients are reluctant to discuss emotional issues with health professionals 76–78. Patients make their own decisions when they are ready to disclose 79. In a study of older adults in Melbourne, Australia, O'Connor et al. found 48% of persons with an ICD-10 moderate or severe depressive episode had not reported any current complaints to their doctor at the time of interview 80. In the MAGPIE study 37% of patients with current psychological symptoms did not disclose their psychological problems spontaneously. Predictors were belief that the GP is not the ‘right’ person to talk to (33.8%) or that mental health problems should not be discussed at all (27.6%). Other reasons include psychological embarrassment, hesitation to trouble the GP, poor clinician–patient relationship, younger age and male gender 80–84. Trust has been directly linked with recognition. Verhaak et al. showed that patients whose diagnosis was recognised tended to express more confidence and trust in their GP 85.
Trust may be an even more important factor for cancer patients, and because there is often such a high level of trust, patients are likely to be accepting of depression treatment in the oncology setting 24. The key lesson for cancer providers is that they should ensure they minimise barriers to care and continue to promote factors that increase trust 24. They should also be sure to provide patients with opportunities to disclose emotional issues.
Lesson 7. Greater contact influences detection
One relatively consistent predictor of diagnostic sensitivity (recognition) is greater contact between provider and patient 42, 85. This can be measured as frequency of appointments, cumulative number of appointments and duration of appointments. Menchetti et al. recently found that those who rarely attended the practice were 2.3 times less likely to receive a diagnosis of depression, compared with those having a high frequency of visits (95% CI = 1.2–4.6) 86. Depressed patients with less than one visit per month were detected 65.3% of the time compared with 83.2% in those with more than one visit per month. In the MAGPIE study from New Zealand, 80.2% of cases seen five or more times during the previous year were correctly identified, compared with 28.8% of those patients not seen in the previous year 29. Cumulatively, only 30% remain undetected at 1 year and 14% at the end of 3 years 87, 88. Generally short appointments are linked with low detection of complex cases and typical consultations are 15–20 min in primary care and cancer care 89, 90. Clinicians often report that lack of time is particularly pressing when dealing with mental health concerns 91. Objective observation supports this concern, showing such consultations are generally more complex and time consuming 91–94. Yet on direct observation a typical mental health discussion can last as little as 2 min 95. Despite this paradox, it is not clear that simply lengthening the consultation improves recognition 92, 96. Good clinicians under pressure may already be attempting to utilise time flexibly in responding to patients' variable needs 97.
The key lesson for cancer staff is that patients must be given enough time with key health-care professionals to express their needs. Also, the fact that cancer patients tend to have frequent visits bodes well for timely detection of emotional problems. If time is a prohibitive factor for clinicians in terms of psychosocial issues then screening can be conducted in the waiting room or other trusted staff could be involved in this aspect of patient care. Nurses, psychologists and social workers can easily be trained to work collaboratively with the cancer physician within the oncology setting to treat patients' depression.
Lesson 8. Clinician confidence/skills influence screening success and subsequent action
There is a wide variation in the ability of GPs to diagnose mental health problems, due in part to differences in knowledge, skills, and attitudes 98, 99. More experienced, confident, outgoing physicians with high academic ability appear to make more accurate diagnoses 30, 100. Yet only about half of GPs feel confident in diagnosing depression or assessing suicide risk 101–106. In a sample of 398 GPs in the Canadian province of Quebec lack of expertise in mental health was associated with a lower incidence of GPs taking on patients with mental health problems 107. Physicians with more positive attitudes to psychosocial aspects of patient care had more psychosocial discussions in visits 108. Doctor-related factors influencing recognition include inadequate interview and diagnostic skills 109, 110, insufficient undergraduate and postgraduate training 111, insufficient time devoted to adequate diagnostic assessment, and a lack of acquisition of new knowledge relevant to provision of treatments. Service-related factors include insufficient remuneration for psychological interventions, insufficient support from specialist public and private mental health services, and inadequate access to non-medical mental health professionals. Doctor characteristics associated with willingness to assess were being aged over 35 years, having an interest in mental health, having had previous mental health training, being in part-time practice, seeing fewer than 100 patients per week, and working in regional centres 39. Much research has examined detailed communication influences on identification and treatment. For example, if a GP adopts a more patient-centred style, this will lead to the patient offering more cues 109 and improved recognition 112. Other aspects of GPs' verbal communication that might be influential are offering more cues, asking questions about psychological issues and eye contact 113, 114. These primary-care studies show that clinician factors influence identification and delivery of care.
In cancer settings there is already a strong track record in relation to communication skills 115, 116. We suggest further research examines other clinician attributes such as confidence, diagnostic skills and therapeutic ability. Additionally, early depression training experiences should be developed as a part of the cancer education programme to increase abilities and confidence early on. Those who still feel uncomfortable can work with other multidisciplinary team members to help with this important aspect of care.
Lesson 9. Training may improve confidence but effects of long-term outcomes are modest
Education and training in primary care has been very effective in improving consultation style and modestly effective in improving recognition, but it remains uncertain whether this translates into improved clinical outcomes 117–119. Educational interventions can be a sole intervention or may be embedded within a package of care that may include organisational changes such as collaborative care or nurse care management. In the Hampshire Depression Project from England, practice teams underwent 4 h of learning about depression in addition to tailored materials. Despite 80% of participants believing that the education would change their management of people with depression, the results displayed no significant differences in rates of diagnosis of depression, referral to specialist psychiatric services or prescription of antidepressants between educated and control groups 120. Gilbody et al. conducted a meta-analysis of 36 studies focusing on educational and organisational interventions for depression in primary care 121. They found collaborative care models that allow shared care between primary care, psychiatrist and psychologist to be effective. They also found positive outcomes in studies with case management interventions, where there was patient education, ongoing support and monitoring. Studies that incorporated physician education or guideline implementation as one part of a multifaceted approach to recognition and management were found to be effective, but educational interventions or guideline implementation in isolation had minimal impact. A meta-analysis by Christensen et al. of 55 studies included 20 focusing on guideline implementation and 17 with educational training interventions 122. Their findings were that neither guideline implementation nor general educational training resulted in any significant improvement of outcomes.
Of course the influence of training on outcomes in oncology settings has already been studied in relation to communication skills but very few educational interventions have been conducted outside of this area 123. More cancer clinicians should be trained in the assessment and management of mood disorders, especially if this training is in the context of a multi-faceted programmatic approach to improving depression care.
Lesson 10. Screening is generally ineffective without aftercare
Early studies were clear that improvements in detection would lead to improvements in treatment (and quality of care) 25, 124–128. Against this, several trials incorporating feedback of scores failed to find any benefit 129, 130. In one of the only well-designed RCT for detection of distress, Hoeper and colleagues found that feedback of a positive GHQ score had no effect on the detection of distress among 1469 patients in a Wisconsin primary-care office 2. More recent trials have focussed on depression. For example Christensen et al.131, 132 recruited 38 GPs in Aarhus County, Denmark and for 1785 consecutive patients randomised 900 questionnaire results (using the SCL-8) to be fed back and 885 to remain blinded. The recognition rate for depression was 35%. Disclosure of the screening questionnaire increased GPs' recognition of mental disorders by 6.6% for patients screened positive, with the greatest effect in those GPs with moderate or low recognition rates. Conversation on psychological topics and rates of planned follow-up consultations also modestly improved. In primary care seven studies have randomised GPs to receive or not receive screening (and screening results) using depression scales 133. Gilbody et al. found that two-step screening with feedback improved recognition especially in high-risk settings. Factors that reduce the benefits of screening include not providing clinicians with interpretation of scores, no mandating follow-up, not linking screening with training or other types of clinician support 134.
The key lesson for oncology settings is that screening alone is not sufficient and screening implementation must combine screening with resources for further care. Further work is necessary to optimise aftercare and delineate acceptable and efficient treatment and this may need to be customised to each health-care setting.
This overview of screening research in primary care reveals many important lessons for cancer settings. However, researchers working in cancer care have also been at the forefront of several important innovations not least is the focus on broadly defined mental health issues and not depression alone 135. Screening for depression or even depression and anxiety is probably not sufficient to capture all emotional distress 136. Primary care is an important partner in mental health services as well as in cancer services 137, 138. About a quarter of people with cancer talk with their GP in the last year about emotional problems 139. Hence, there is scope to develop psychosocial aspects of cancer care in primary care 140. Although there are many variations in health-care systems of delivery, important lessons emerge. We acknowledge that there is no perfect method of diagnosing depression, indeed the criterion standard may be improved or simplified with DSMV and ICD11. However, most non-psychiatric clinicians have difficulty even recalling the current criteria for major depression 141. Further, only one-third claim to make diagnoses based on validated criteria 142. Yet recent research on distress in primary care shows that clinicians also struggle to identify emotional distress with success in perhaps one in three individuals 143. This suggests that barriers to identification are greater than just difficulty with the current criteria for depression. In primary care, time and resources are limited and hence psychological treatment and often structured self-help programmes are not always available. Inadequate treatment is well documented and the most plausible factor explaining under-treatment is under-recognition. Yet historically there has been an undue emphasis on antidepressants as first-line treatment. Antidepressants are typically the treatment of choice for physicians, but not for patients and hence first-line treatment for managing depression can create a mismatch between clinician and patient opinion 144. This can lead to understandably high rates of discontinuation 145.
Given that identification of mental health concerns is key, two major factors appear to influence detection. These are how the person with depression (or distress) describes their symptoms, and how the clinician interviews the patient. The nature of the therapeutic relationship underlies both of these factors. Even in the face of a high frequency of contact a therapeutic relationship, which is noted by the clinician (or patient) to be unhelpful is likely to decrease recognition rate. Discussion of emotional distress in primary care is also linked with high patient satisfaction 146. Additional factors such as the skill of the clinician and the use of tools may also play a role. There are certainly many potential barriers to successful diagnosis and treatment. Mental health skills training has been effective in improving recognition and management of somatising and depressed patients by GPs, but it remains uncertain whether this translates into improved clinical outcomes 117, 119. Interventions are likely to be most successful where problems are most serious. For example Shapiro et al. conducted a randomised clinical trial involving 1242 patients attending inner-city primary care by giving feedback of GHQ scores. Results showed marked increases in detection, but only among the elderly, black African Americans, and men 147.
Clearly there is a wide variation in the ability of GPs to diagnose mental health problems, due in part to differences in knowledge, skills, and attitudes 98, 142, 148. There are also variations in management and referral of complex cases 149, 150. Screening has the effect of reducing some of the heterogeneity in identification, but at a cost of additional time for all those involved. Primary-care screening and associated care is generally cost-effective, but this has not yet been studied in cancer setting 151. Most questionnaires are usually acceptable in primary care for initial introduction in a trial 152, but only short, simple questionnaires retain acceptability for multiple applications. Evidence-based care appears to be predicted by stronger confidence in depression identification, less perceived time limitations, and less perceived barriers for guideline implementation 153.
There are several limitations of this focussed narrative review. We only reviewed English language publications, and thus may have omitted key works that represent other cultures. Additionally, we did not use formal meta-analytic techniques due to the number and complexity of the issues we were considering. Nearly 40 years of research have revealed key lessons for screening and detection of mood disorders in primary care. Many of these lessons apply directly to cancer settings although there are also differences. In primary care often the physician is most involved with depression care, whereas in cancer settings it is often the clinical nurse specialists who are most hands-on. In primary care there is a complex collection of physical diseases, whereas in cancer care there is more homogeneity. Depression in primary care increases the likelihood of primary-care attendance, but this may not be the case in cancer care 154. In primary care there is a large proportion with acute and minor medical conditions, but in cancer care there is a significant proportion in the late stage of life. However, there is good reason to believe that some of the differences in settings may result in easier implementation and acceptance of depression screening programmes in oncology settings than in primary-care settings. Oncologists already recognise the emotional toll that cancer takes on patients, and have established partnerships for psycho-social care. Additionally, cancer providers understand the symptoms of cancer and treatment side effects. With some training on depression and related mood disorders they are well positioned to understand true symptoms of depression. There is typically a high degree of sustained contact between cancer patients and their providers, leading to a high degree of trust, which is likely to influence patients to accept both depression screening and treatment. Last, because of the sustained contact, aftercare and follow-up are facilitated, and patients should not ‘fall between the cracks’. In all, we believe that taking into account the oncology context, the lessons learned from four decades of research on depression screening in primary care may help enhance clinical practice and improve research goals for those working in psychosocial oncology.