Validity of Self-Reports About Quality of Life Among Patients With Schizophrenia

Quality-of-life measures are used routinely in assessing clinical outcomes in both the treatment of acute care patients and the rehabilitation of chronic care patients. Objective measures of quality of life are those that can be reliably measured and are easily observable and quantifiable, such as income, years of education, and employment status. Subjective measures of quality of life are self-reported measures of satisfaction with objective factors, such as how satisfied one is with one's job.

A problem arises when objective and subjective measures of quality of life show a discrepancy. Should we attach more weight to one or the other? The problem is compounded when the individuals whose quality of life is being assessed have been diagnosed as having schizophrenia. Because they may lack insight into their illness and because their mental functioning may be impaired, the credibility of their self-reports is questionable (1).

One way to assess quality of life in clinical populations is by using a "knowledgeable other person" or proxy—someone who reports on behalf of another. The purpose of using proxies in quality-of-life research with psychiatric patients is to validate patient responses by comparing proxy and patient responses on similar measures. In addition, from a practical perspective, if proxies' responses are found to be substantially correlated with patients' responses on quality-of-life measures, then a proxy response may be substituted for a patient's response in studies using quality-of-life measures whenever the patient is unavailable or is not in a fit state to be interviewed.

Methods

In this study we measured the subjective and objective aspects of quality of life among patients with schizophrenia using reports of the patients themselves and the reports of caretaker-proxies. We included a control group of nonpsychiatric chronic medical patients with cancer and their caretaker-proxies so that the concordance rate between the patients and their proxies could be compared between the two groups.

Seventy-four people were recruited for the study in 1997 and 1998. The schizophrenia group, recruited from an outpatient population, comprised 22 patients and 22 proxies. The diagnosis of schizophrenia was made formally by psychiatrists using DSM-IV criteria. The cancer group, also recruited from an outpatient population, comprised 15 patients and 15 proxies. The patients with cancer had received their diagnoses from oncologists. All patients in both groups had received their diagnosis at least one year before the study, so enough time had passed for their illness to have an impact on their quality of life.

The schizophrenia group had 17 men and five women, and the cancer group had eight men and seven women. The mean±SD age of the schizophrenia group was 34.7±8.9 years, and that of the cancer group, 31.3±4.8 years. The schizophrenia group had a mean±SD of 11.7±1.8 years of schooling, and the cancer group, 15.4±2.2 years.

Proxy status was determined by the patients themselves; they designated one of their parents or a spouse as their proxy. Eighty-six percent of the patients with schizophrenia chose parents as proxies, and 80 percent of the cancer patients chose spouses. Only 18 percent of the patients with schizophrenia were married, compared with 80 percent of the cancer patients. Similarly, 18 percent of the schizophrenia group were employed full-time, compared with 87 percent of the cancer group.

The patients with schizophrenia were screened to ensure that they had no history of cancer, and the patients with cancer were screened for any history of psychiatric diagnosis and treatment. All patients and their proxies gave written informed consent to participate in the study.

The patients and their proxies were given Lehman's Quality of Life Interview (2) individually. The patients were asked to respond to the items from their own perspective, and the proxies were asked to respond as though they were the patient. The same person interviewed all participants. Interrater reliability was verified by using a second rater who independently rated eight participants, either a patient or a proxy, from behind a one-way mirror. Agreement between the two raters was 100 percent.

Results

Chi square analysis revealed no significant gender difference between patient groups. However, a significant difference was observed between proxy type—that is, the relationship between patient and proxy (χ²=16.29, df=1, p<.001). The patient with schizophrenia was more likely to choose a parent, and the patient with cancer was more likely to choose a spouse. As expected, there was also a difference between the patient groups in marital status (χ²=16.88, df=1, p< .001) and employment status (χ²= 16.88, df=1, p<.001). Student's t tests revealed no significant difference between the groups in age, but a significant difference was observed in level of education (t=5.61, df=35, p<.001).

The correlation between the global objective and subjective quality-of-life indexes was negligible and nonsignificant for patients with schizophrenia but significant for patients with cancer (r=.57, p<.05). A nonsignificant correlation was observed between these scores for the schizophrenia proxies, and a significant correlation was noted between scores for the cancer proxies (r=.58, p<.05). The difference between the two correlations was not significant for the patient groups, but it was significant for the proxy groups (z=2.85, p<.01).

The concordance between patient and proxy responses in the schizophrenia group was significant on objective scores (r=.72, p<.001), but not on subjective scores. The concordance between patient and proxy responses in the cancer group was significant on objective scores (r=.69, p=.005), but once again, not on subjective scores. The difference between the concordances was not significant for either group.

Discussion and conclusions

The patients with schizophrenia differed from those with cancer in marital status, proxy type, education, and employment, with the latter more likely to be married, to have a spouse as a proxy, to have more schooling, and to be employed. However, we see no reason why these demographic differences would have affected the main results or their interpretation.

The correlation between objective and subjective quality-of-life ratings was significant for the cancer group but negligible for the schizophrenia group. This result confirms findings from previous studies (1,3,4) showing a greater difference between objective and subjective quality of life for patients with schizophrenia than for patients with cancer.

However, we also observed a greater difference between objective and subjective quality of life for the proxies of patients with schizophrenia than for the proxies of patients with cancer. The finding that the ratings of proxies of patients with schizophrenia produce the same kind of discrepancy as those of patients with schizophrenia suggests that the reported difference between objective and subjective quality of life is a valid one. Perhaps patients with schizophrenia eventually adapt to their reduced circumstances and lower their expectations accordingly over time, eventually becoming satisfied with less. Whatever the explanation, the findings suggest that the self-reports of patients with schizophrenia can be taken at face value and that for patients who are too disturbed to cooperate, a caretaker-proxy can answer accurately on their behalf.

As expected, greater concordance was found between patients and proxies for objective indexes than for subjective indexes. This result likely occurred because objective indexes are more reliable—that is, they can be measured more accurately.

A major limitation of this study was its small sample. Future research should use larger samples to increase the power of the statistical tests. In addition, although confidence in patients' diagnoses was not an issue, we would have had even more confidence if individual diagnoses had been validated independently by other psychiatrists and oncologists and standard exclusion screens had been used to identify patients with psychiatric and cancer histories.

Ms. Khatri is a doctoral student and Dr. Romney is a professor in the department of psychology at the University of Calgary, in Calgary, Alberta, Canada. Dr. Pelletier is a clinical psychologist in the department of psychosocial resources at the Tom Baker Cancer Center in Calgary. Send correspondence to Dr. Romney, Department of Psychology, University of Calgary, 2500 University Drive N.W., Calgary, Alberta T2N 1N4, Canada (e-mail, [email protected]).