Close association was found between patient's self-report and investigator's rating of SDI and SD, although investigator ratings were lower than patients. The fact that social workers underestimate SD reported by patients is perhaps not surprising. Social workers tend to have people referred to them who are in extremis
with complex family or financial problems. This means their ‘benchmark' may be at a higher starting point than the average patient. In addition, they may pay more attention to problems for which they have a remedial action. Other health-care professionals, for example, nurses or doctors, may have had a different interpretation of the social issues elicited during interview. In a study comparing quality of life (QL) scores over 12 domains, physicians underestimated the severity of patient's experience in 10 of the symptom/functioning scales (Petersen et al, 2006
). Of the two scales overestimated by physicians, the social functioning scale showed greatest disagreement, possibly reflecting the area physicians tend to have least experience of from their clinical practice.
Using novel methodology, a cutoff for SD was derived, an initial algorithm was developed and guidance for using SDI in clinical practice produced.
For patient-centred assessment to have clinical value, instruments must be evaluated beyond basic psychometric properties. This has been achieved in psychiatry. Instruments such as the HADS have ‘cutoff' levels for case level anxiety or depression detection using DSM systems derived from interviews (Razavi et al, 1990
). There is no equivalent system for calibrating social difficulties, and therefore the decision to take the top 10% of SDI scores, as identified by the social work investigator, was made as the best available ‘gold standard'. This is a study limitation made for pragmatic reasons. However, the area under the ROC curve, derived using this methodology was 0.85, a level generally thought of as being in excess of acceptable levels required for screening and validated by good inter-rater reliability demonstrated between social work investigator and oncology social worker.
Rasch analysis of the SDI has provided the basis to derive a cut-point using this ‘gold standard'. As there is no differential item functioning within the Rasch SD scale for age (Smith et al, 2007
), the fact that older patients were more likely to refuse to participate should not have influenced the ‘cut-point' calculated. The data were collected from only one cancer centre. Although provision of support services and the level of expertise of clinical staff will vary across the cancer services of UK, this is unlikely to alter the cut-point derived from Rasch due to the lack of differential item functioning. As this work progresses, we will be able to test whether item invariance holds. How health-care professionals choose to interpret the guidelines within their own settings may well be influenced by local service provision and training of staff.
There are a number of ways of utilising patient self-reported questionnaires in everyday practice. The CARES-SF has normative data available for comparative purposes. In addition, it provides the patient with the option of stating whether they would like help with items, although documentation on how this relates to item scores is limited (Coscarelli and Heinrich, 1988
). Recommendation for use in clinical practice is for preliminary assessment by CARES followed by a brief interview. The Distress Thermometer provides a quick assessment of general distress with the potential for problematic items from a number of domains to be flagged (Roth et al, 1998
; Jacobsen et al, 2005
). Recent guidance recommends that people with a score of 4 or more should have a clinical assessment by the primary oncology team (National Comprehensive Cancer Network, 2007
). Patients with flagged problems may be overlooked if their distress score falls below 4. This is particularly relevant to problems concerning childcare, insurance, transportation and work, which were not associated with reported distress (Jacobsen et al, 2005
If an SD cutoff of 10 or more had been employed as the only guide for further discussion by the clinical team, 72 patients with item rates of ‘2' or ‘3' would have been missed of whom 30 were either referred for specialist help or provided with information. The SDI was designed to provide additional information for doctors and nurses, to highlight issues of concern and to enhance the health-care professional/patient consultations with a view to identifying patients who may benefit from support, advice or intervention. Although the SD score does provide a cut-point indicating more severe levels of SD, this should not be taken as an automatic reason for referral to social work. There will be considerable differences between patients in terms of what each regards as a severe difficulty depending upon the individuals' personal situation; for example, not being able to get out of the house for one person may be a great restriction, whereas for someone else may be of little consequence. The SDI output is a starting point for discussion and, as with any measure, SDI scores should not be employed in isolation. Decisions to intervene, even with established tests such as X-rays, based on X-ray alone would be foolhardy as other clinical and social issues may be influential. Definitive decisions using only SDI scores would result in clinics being overwhelmed by large numbers of identified patients and staff feeling reluctant to engage with the assessment process. A balance must be found between assessment frequency, SDI results and other clinical and sociodemographic considerations to keep patient and staff burden to a minimum without losing sensitivity of identifying patients who are struggling. Earlier work has shown that younger people and those with advanced disease are particularly vulnerable to the social impact of cancer (Wright et al, 2002
). Although sociodemographic and clinical information may not be used as a proxy for identification of social difficulties, it may be the case that clinics in which more vulnerable groups of patients attend should be prepared to have to respond to a higher level of need. This will not only require good staff training but also access to up- to -date and relevant information on services and support.
Referral rate to social work and psychology was increased three-fold in this study compared to standard practice, although the majority of participants did not require intervention. The researcher was a social worker with many years of oncology experience in the Yorkshire region. Not only did the research interview provide adequate time for eliciting problems from patients but also the interviewer was knowledgeable and confident about local support services and information available. Prior to the start of the study, discussion had taken place with the psychosocial oncology service and the oncology social work service that had both offered active support for the study. Some clinical staff may feel that they would rather not get involved in these types of discussion due to limited communication skills (Fallowfield and Jenkins, 1999
) and also a lack of knowledge of local resources. Often, clinics are busy and lack privacy; staff have to undertake a number of tasks simultaneously, meaning that psychosocial assessment and referral may be neglected. Of those who received either information or referral, many could have ‘helped themselves' if information had been available in clinic, for example, holiday and disabled parking permit information. A number of patients not experiencing problems were referred to social work for welfare benefits advice. These people, entitled to welfare benefits but unaware of their entitlement, may not have been picked up by the SDI. Again, good information displayed in outpatient clinics may overcome this inconsistency. The SDI may provide the means to identify people experiencing difficulties sooner with a simple advice from clinic staff providing easy resolution resulting in fewer patients developing complex problems requiring referral.
Not all problems require interventions; possibly a simple acknowledgement of or reference to the patient's situation may be sufficient to enhance well-being. In a study in which QL information completed by patients in a three-armed randomised trial, chronic nonspecific symptoms were discussed more frequently in consultations where patients had completed the QL assessment and this had been fed back to the physician in real time. This did not result in either longer consultations or change in patient management but did lead to a significant improvement in QL and emotional functioning (Velikova et al, 2004
). There were a significant minority of participants who were struggling with one or more aspects of their everyday lives, who had minimal support and had not been identified in standard clinical practice. It is for these people, the SDI may play a role with the opportunity to discuss ‘flagged' difficulties or overall SD with the clinical team leading to possible intervention. Not all interventions are complex and a simple solution may make a big difference to a patient and not be too burdensome for staff.
Having developed guidance for health-care professionals, the next step is to consider who may be best placed to respond to the SDI in clinics. Health-care teams are multidisciplinary with members having different roles and responsibilities (Matthews et al, 2004
; Catt et al, 2005
). It would be unrealistic to expect all team members to have expertise in responding to all items listed within the SDI. An ongoing interview study in which staff and patients are being asked about these issues should provide useful information concerning current levels of knowledge on support services, roles, responsibilities and expectations. Once this has been established, team training on score interpretation, agreed management and support services will be developed.
Future work will focus on whether or not this type of assessment applied routinely will lead to a change in management of patients or an improvement in patient's well-being.