The current study was a collaborative project between the University of Hong Kong, the Hospital Authority, and the Chinese University of Hong Kong.
Nasopharyngeal carcinoma (NPC) is highly prevalent in southern China. Prominent acute side effects of radiotherapy create problems in daily living and working that can generate considerable financial difficulties. A better adjustment to a diagnosis of NPC appears to be associated with an improved rate of recovery, a better quality of life (QoL), a quicker return to work, and normal functioning. Patient satisfaction with physician consultation and the way information is provided in particular may have significant bearing on QoL. The current study reports on short-term QoL after radiotherapy in NPC patients as a function of satisfaction with the information provided.
Newly referred Hong Kong Chinese NPC patients (n = 211) completed interview measures at baseline before the initiation of radiotherapy, at 4 months after baseline (immediate posttreatment consultation) (FU 1), and again at 8 months (short-term postradiation period) after baseline (FU 2). Satisfaction with the information provided was measured by five items selected from the cognitive subscale of the Medical Interview Satisfaction Scale (MISS). QoL was measured by the Chinese version of the Functional Assessment of Cancer Therapy-General Scale (FACT-G (Ch)).
After adjustment for overall patient satisfaction (the PSQ-9), optimism, worry about family, anger, eating ability, subjective health, family income, and occupation at FU 1, treatment between baseline and FU 1, and disease recurrence after baseline, the 5-item MISS at FU 1 (β = 0.21, P < 0.01) was found to significantly predict patient QoL at FU 2. Adjustment for baseline QoL and disease stage did not appear to alter this relation (β = 0.20, P < 0.01).
Nasopharyngeal carcinoma (NPC) is highly prevalent among populations originating in southern China.1 In 1996, 1177 new cases of NPC were identified in Hong Kong, giving an estimated incidence of 20 cases per 100,000 persons per year, making NPC the fifth most prevalent malignancy in Hong Kong,2 with a mortality rate of approximately 6 in 100,000 persons per year. This finding implies that patients with NPC have a good prognosis.
NPC patients primarily are treated with radiotherapy. Both the acute and long-term side effects of radiotherapy are prominent.3, 4 Difficulties with work shortly after the initiation of radiotherapy can create financial problems for many NPC patients in a country that does not have a welfare system, such as Hong Kong.5 Better adjustment shortly after radiotherapy may be important for NPC survivors, leading to a quicker rehabilitation. Quality of life (QoL) is a comprehensive adjustment index that is being used increasingly in treatment-related oncology studies,6–9 including, at a minimum, physical, functional, psychological, and social well-being dimensions.10
Support has been identified to be an important coping resource11 that is related to positive adjustment in NPC patients.12 Ma reported that NPC patients in Hong Kong consistently sought support from health professionals.13 Broadhead and Kaplan regarded physician informational support, rather than emotional aid, to be the major type of social support cancer patients need during the posttreatment period.14 Patient satisfaction with hospital services in general is associated with subsequent QoL15 and patient satisfaction with information provided by physicians during clinical consultations has been reported to have an important influence on subsequent adjustment.16 High satisfaction with information provided by physicians most likely reflects good patient understanding, thereby facilitating improved information recall.17, 18
A search through MEDLINE Express and PsycLit from 1966–2000 found no studies regarding satisfaction with postradiotherapy consultation and subsequent QoL in NPC patients. Studies regarding other groups of cancer patients have yielded contradictory findings. Neuling and Winefield reported anxiety and depression were significantly related to support from surgeons at 1 month and 3 months after surgery in 58 breast carcinoma patients.19 Butow et al. reported that 142 cancer patients whose questions were answered during their first consultations with physicians were found to have better psychologic adjustment 3 weeks after the consultations,16 whereas Mathieson et al. found family physician support predicted the psychologic state among 45 patients with head and neck carcinoma who were attending follow-up appointments. In contrast, Rutter et al. found expected or perceived satisfaction immediately after consultation did not appear to predict QoL 6 weeks later in 50 cancer patients with mixed diagnoses.21 However, we believe that the small sample sizes used in these studies was a concern and the majority of the studies also were limited by a failure to adjust for medical or other important psychosocial variables.
Other potential influences on patient QoL include optimism and emotional state. Optimism frequently has been found to be related to QoL or adjustment,22–24 although to our knowledge it rarely has been studied among cancer patients. Cheng reported that Hong Kong Chinese NPC patients commonly experienced anger and family-related worry, which may affect postradiation QoL.3
To our knowledge, it is not known whether satisfaction with the information given during a consultation is an important determinant of postradiation QoL in patients with NPC. Consequently, the current study examined whether satisfaction with the information provided by physicians during consultation immediately after the completion of radiotherapy predicted QoL 4 months later in a substantial sample of NPC patients after adjustment for medical, demographic, and psychosocial variables.
MATERIALS AND METHODS
Participants were newly referred Chinese NPC patients from the outpatient sections of the Departments of Clinical Oncology in five regional hospitals in Hong Kong, China, recruited into a larger psychooncologic QoL study between 1996–1997, and followed between 1996–1998. All the patients were ages 18–85 years and had provided verbal or written informed consent for participation. Two hundred patients were targeted to maintain a sample size that was sufficiently large enough to minimize Type II errors.25
Design and Sampling
Data were collected at baseline before radiotherapy, and subsequently at Follow-up 1 (FU 1) and Follow-up 2 (FU 2). Radiotherapy in the current study refers to the first course of radiotherapy of 6 weeks' duration, unless otherwise stated. FU 1 was conducted at the hospital immediately after the physician-patient consultations at the completion of radiotherapy (around 4 months after referral), or by telephone shortly after these consultations. FU 2 occurred 4 months after FU 1.
Data were collected simultaneously in the five hospitals covering both urban and rural areas of the territory. Patients were recruited during time slots for clinical consultations, visits to the radiotherapy planning section, and visits to receive radiotherapy.
Depending on manpower and patient availability in each targeted clinical session, either consecutive or systematic sampling was employed. In consecutive sampling, every patient (ages 18–85 years) scheduled for a clinical visit in the targeted time slot was approached in turn and invited to participate into the study. In systematic sampling, a 1 in 5 or 1 in 10 sampling method was adopted when large patient numbers or low interviewer availability predominated.
All data were collected by interview using a standardized questionnaire. Face-to-face interviews were performed whenever possible but telephone interviews were necessary in both follow-ups due to manpower limitations. The informed consent for participation was obtained from the participants. Interviewers read out the questions one by one and recorded the responses of the patients. Minimal probing was used whenever necessary.
All interviewers were qualified social workers who were trained intensively before they conducted the interviews to facilitate interviewing consistency. Definitions of terms in single items were conveyed to the interviewers to ensure as much consistency as possible in the presentation of these measures. Periodic inter-rater reliability remained at > 0.9.
Patient consultation satisfaction
The Medical Interview Satisfaction Scale (MISS)26 is a 26-item, 5-point Likert-scored instrument comprised of cognitive, affective, and behavioral subscales with 5 response alternatives labeled “Strongly agree,” “Agree,” “Uncertain,” “Disagree,” and “Strongly disagree.” The nine-item cognitive subscale measures patients' perceptions of the adequacy of the physicians' explanations and ability to provide information and the patient's understanding of the medical information.26 The internal consistency (Cronbach alpha [α]) of the MISS is 0.9.26 The MISS has been reported to be related to patients' need fulfillment, with patients who had a greater number of their expectations regarding consultation met reporting higher consultation satisfaction compared with patients who had fewer expectations met.27 Patients' satisfaction with the way in which physicians provided information was measured by five items of the cognitive subscale. These items were selected based on their relevance to NPC patients. They were translated into Chinese for use in the current study (Fig. 1).
All items were double back-translated by a team of language specialists, health care professionals, and social workers. The drafted items were piloted on five randomly selected patients to identify any refinements that were needed. The translation was reviewed further and recommendations were made until acceptable wording was agreed on.
Short-term postradiotherapy QoL
QoL was measured by the Chinese version of the Functional Assessment of Cancer Therapy-General Scale (FACT-G (Ch)).28 The 29 items of the instrument are scored as 0 = “Not at all”, 1 = “A little bit”, 2 = “Somewhat”, 3 = “Quite a lot”, and 4 = “Very much”. The FACT-G (Ch) has five subscales, measuring physical, social/family, emotional, and functional well-being and relationship with physicians. The FACT-G (Ch) has good psychometrics and acceptable cultural equivalence.28
Psychosocial variables for adjustment of the results
Patient satisfaction with hospital services was measured with the Patient Satisfaction Scale (nine-item version) (PSQ-9).29 The scale was commissioned, developed, and validated locally to measure patient satisfaction with specialist outpatient services provided in Hong Kong. Responses are on a five-point Likert scale with response categories labeled “Very satisfied”, “Satisfied”, “OK”, “Dissatisfied”, and “Very dissatisfied”. The PSQ-9 is a carefully developed indigenous instrument with good psychometric properties.29
Visual analogue measures using single-item, 10-cm lines marked “0” and “10” and scored continuously were used to assess optimism and affect. Optimism was measured by one item that stated “My attitude towards life is”…, which was headed “completely pessimistic” “0” and “completely optimistic” “10”. Emotional responses to stress are shaped by appraisals of coping resources11 with optimism reflecting a disposition that tends to anticipate better outcomes. Optimism therefore is expected to be associated with more positive affect, suggesting the constructs “optimism” and “affect” may be somewhat related conceptually. The item measuring optimism correlated moderately with another item addressing depression (Pearson correlation coefficient [r] = -0.4).30
Two affect measures were used. Worry about family (Worry) was measured by an item in the form of a statement “Regarding the future of my family/family members, I am….” The “0” end is headed “not worried at all” and the “10” end is headed by “very much worried”. Anger was measured by one item, “I feel angry”, rated on a five-point Likert scale of “Not at all”, “A little bit”, “Somewhat”, “Quite a lot”, or “Very much”. The 2 items had low correlation (< 0.18) with dissimilar social desirability measures, but was found to be correlated moderately (r = 0.3 and P < 0.001 for Worry and r = 0.4 and P < 0.001 for Anger) with an item addressing depression, indicating reasonable divergent and construct validity as measures of negative affect.30
Demographic and medical data
Categoric demographic data were collected during interviews concerning age, gender, education, marital status, religion, occupation, and family income. Main medical data were traced through participants' hospital medical records. This included disease stage, using the Ho staging classification31 (Fig. 2); treatment administered between baseline and FU 1 (radiotherapy/combined treatment) and between FU 1 and FU 2 (no/yes); course of radiotherapy received between baseline and FU 1 (> 1 course/1 course); duration of first radiotherapy course (days); and subsequent recurrence after baseline (no/yes) (cutoff point was 1 month after FU 2).
Other medical data were collected at the time of the interviews. Pain was measured by an 11-point item selected from the Wisconsin Brief Pain Questionnaire (BPQ),32 asking “How much pain do you have right now?” The “0” end headed “no pain” whereas the “10” end headed “pain as bad as you can imagine.” Current perceived health state (CPH) was assessed using the question “How would you rate your health right now?” and was rated on a four-point scale rated “very good”, “good”, “poor”, and “very poor”.33 Eating ability was rated by an 11-point (0–10) item in the form of a statement “My eating ability is….” At the “0” end it was headed “very bad”, and at the “10” end it was headed “very good”.
Standard descriptive analyses (mean, standard deviation [SD], and chi-square) assessed the sample characteristics. One-way repeated measures analysis of variance (ANOVA) compared the mean levels of single variables across time. Scree plot-driven factor analysis, α, and r were used to confirm the integrity of the five-item MISS. Factor loadings of < 0.4 were suppressed. Forced entry multiple linear regressions with mean substitution for missing data were used to explore to what extent satisfaction with the information provided predicted short-term postradiation QoL and to adjust for the effects of other factors. Separate regression models initially were run for demographic and medical variables. Variables with a P value < 0.10 were included in the final regression model, which used the MISS at FU 1 as the predictor of FACT-G (Ch) at FU 2. Other variables at FU 1, plus identified demographic and medical variables, were used for adjustment. The normality of scores on the dependent measure was assessed using the Kolmogorov-Smirnov goodness of fit test (Z). Diagnostic procedures using residuals deleted using the Student t test were employed to assess regression model assumptions and adequacy of the model fit. Multicollinearity was excluded by reference to tolerance values and condition indices.25
A total of 470 of 748 eligible new cases (63%) were sampled between 1996 and 1997, 211 of which (45%) were recruited to form the current study sample. The overall refusal rate was low (10%). Nonavailability of interviewers accounted for the largest proportion of nonrecruitment (23%). The number of patients interviewed declined from 211 at baseline to 201 at FU 1 and 187 at FU 2, a follow-up rate of 88%. Of those 24 participants not completing all interviews, 6 had died (3%), 4 refused (2%), 10 were lost to follow-up (4%), and the remaining 4 participants had immigrated (2%). The mean duration between baseline and FU 1 was 3.8 months (SD = 20.2 days) and the mean duration between FU 1 and FU 2 was 3.9 months (SD = 20.6 days). The mean time from the end of radiotherapy to FU 1 was 38 days (SD = 12.0). The short-term postradiotherapy period (the time between the end of radiotherapy and FU 2) was on average 5.2 months (mean = 157 days, SD = 33.5 days). At FU 1, 73% of the interviews were conducted face to face and the remainder (27%) were conducted by telephone interviews. More interviews (53%) were conducted by telephone at FU 2 because of logistic constraints.
Demographic and Medical Profile
The demographic profile of the current study sample at baseline is shown in Table 1. The gender ratio approximated that of the population of Hong Kong Chinese NPC patients. No gender difference was found with regard to age (Student t test = 1.26, degrees of freedom [df] = 20.9; P value = not significant). Approximately 73% of the subjects were ages 30–60 years and in active employment. Approximately 57% of the participants were working full-time at baseline, a number that decreased to 37% at FU 1 and 34% at FU 2 (chi-square test = 30.5, df = 6; P < 0.001). Similarly, the proportion of subjects whose family income was < Hong Kong $10,000 (US$1282) per month increased significantly from 30% at baseline to 41% at FU 1 (chi-square test = 25.7, df = 12; P < 0.05) and then declined slightly to 38% at FU 2.
Table 1. Main Demographic and Medical Data for the Current Study Sample (n = 211)
SD: standard deviation; FU 1: Follow-up 1 (conducted immediately after completion of radiotherapy; FU 2: Follow-up 2 (conducted 4 months after Follow-up 1). Figures are numbers of cases unless otherwise stated.
Percentages of all newly identified nasopharyngeal carcinoma patients in 1994 (Source: Hong Kong Department of Health Annual Report 1997/1998).
$1 U.S. = $7.8 HK.
The item that tapped religion was added after the start of data collection.
Recurrence after baseline indicates a recurrence documented from after baseline to 1 month after the second follow-up.
Eating ability at Follow-up 1 was scored as 0–10, the higher the better.
Duration since first course of radiotherapy indicates the duration from the end of the first course of radiotherapy to the 2nd follow-up.
Pain at FU 1 indicates pain scored at the time of the first follow-up on a scale of 0–10, the higher the score, the worse the pain.
An average of 20 days (SD = 48.0) had elapsed between patient diagnosis and baseline recruitment. In Table 1, 41% of patients had Ho Stage I or II disease, 49% had Stage III disease, and the remainder of the patients had Stage IV or Stage V disease. Twenty-three patients (11%) developed a recurrence. All but one recurrence occurred after baseline but before FU 2. A majority of participants (89%) received radiotherapy only between baseline and FU 1, with 90% receiving a single course of radiotherapy (mean = 47 days [SD = 33.5]) throughout the duration of the project. Only 12% received treatment between FU 1 and FU 2. The mean pain level rose significantly from baseline to FU 1 and then declined by FU 2 (F2,370 = 6.8, P < 0.01). Mean levels of CPH were not found to differ significantly across time (F2,366 = 0.5, P value = not significant). Participants experienced moderate eating difficulties at FU 1 (mean = 6.2 and the median = 6.0 [SD = 2], range, 0–10, with the lower the number the worse the difficulty).
Validation of the MISS
The factor structure, internal consistency, and construct validity of the five-item MISS were explored. Scree plots indicated a three-factor solution was appropriate for a factor analysis of the five-item MISS, the PSQ-9, Optimism, Worry, Anger, and the FACT-G (Ch) items at FU 1. The five-item MISS (factor loading 0.48), the PSQ-9 (0.65), the two items in the Relationship with physician subscale of the FACT-G (Ch) (0.54 and 0.61, respectively), and Optimism (0.45) loaded on the same factor, indicating that the translated MISS items were similar conceptually to these items concerning satisfaction with health care professionals.
The MISS items had acceptable internal consistency (α = 0.72). Access to physicians and hence information in Hong Kong generally is more restricted than in the West. Logically, this appears likely to affect patient satisfaction with the giving of information and thus we expected satisfaction and availability to be related. However, when the FACT-G (Ch) Relationship with physician subscale item “My doctor is available to answer my questions” was used to explore the construct validity with the 5-item MISS at FU 1, the 2 were only modestly correlated with each other at 0.3, indicating < 10% of shared variance.
Mean and SD of the MISS, the FACT-G (Ch), and Other Psychosocial Measures
The mean and SD of the five-item MISS, the FACT-G (Ch) and other psychosocial measures are shown in Table 2. The FACT-G (Ch) at FU 2 approximated normality (Kolmogorov-Smirnov Z test = 1.2, P value = not significant) and therefore no transformation of the variable was deemed necessary. Significant changes in means across time were observed for the FACT-G (Ch), the PSQ-9, Anger, and Worry. Mean MISS scores did not appear to differ significantly over time. Although the FACT-G (Ch) and Anger scores increased progressively, Worry scores appeared to decline with time. The PSQ-9 dropped at FU 1 but then recovered. Post hoc Bonferroni multiple comparisons revealed significant differences only between the baseline and FU 1 means, but not between FU 1 and FU 2 for the FACT-G (Ch), the PSQ-9, Anger, and Worry (P < 0.01).
Table 2. Repeated Measures ANOVA: Mean Comparison of the FACT-G (Ch), the 5-Item MISS, and Other Psychosocial Variables Across Time
Mean (SD) FU 1
ANOVA: analysis of variance; FACT-G (Ch): Functional Assessment of Cancer Therapy General Measure (Chinese version); MISS: The Medical Interview Satisfaction Scale; SD: standard deviation; FU 1: first follow-up; FU 2: second follow-up; PSQ-9: The Patient Satisfaction Questionnaire (Nine-item version); Anger: the higher the score the worse; Worry: Worry about family, the higher the score the worse.
Age, gender, education, marital status, religion, family income, and occupation at FU 1 were included in a forced entry multiple regression model to predict FACT-G (Ch) at FU 2. Only family income (β = 0.12, P = 0.09) and occupation (yes/no) (β = -0.17, P = 0.02) were found to be significant at the 10% level and were retained for further analysis.
Similarly, all medical variables were entered into a separate regression model. Treatment received between baseline and FU 1 (β = -0.24, P = 0.00), eating ability at FU 1 (β = 0.21, P = 0.00), and CPH at FU 1 (β = 0.17, P = 0.01) appeared to predict the FACT-G (Ch) at FU 2. Although recurrence after baseline was not found to be significant at the 0.10 level (β = -0.10, P = 0.15), it was included for further analysis because information regarding recurrence after baseline may impact on the subsequent physical status of the participants.
Information Satisfaction and QoL
The five-item MISS (adjusted for selected psychosocial, demographic, and medical variables) was included in a final forced entry regression to predict the FACT-G (Ch) at FU 2 (Table 3). The 5-item MISS was found to significantly predict FACT-G (Ch) scores (β = 0.21, P < 0.01), indicating that NPC patients reporting more satisfaction with the medical information provided an average of 1 month after the end of radiotherapy had a better QoL 4 months later.
Table 3. Multiple Regression (Forced Entry): The Effect of the 5-Item MISS at FU 1 on QoL at FU 2, Controlled for Other Psychosocial and Selected Demographic and Medical Variables (n = 211)a
Variables were measured at 1st follow-up (FU 1) unless otherwise stated. n.s.: non-significant P value at 0.05 level; Five-item MISS: Five selected items from the Cognitive subscale of the Medical Interview Satisfaction Scale, scored on a scale of 5–25; FU 1: first follow-up, QoL: quality of life; FU 2: second follow-up; SE: standard error; PSQ-9: Patient Satisfaction Questionnaire (nine-item version), on a scale scored 9–45; NS: not significant; Optimism: on a scale scored 0–10; Worry: worry about family, scored on a scale of 0–10; Anger: scored on a scale of 0–4; Treatment (baseline/FU 1): treatment received between baseline and the first follow-up (radiotherapy alone/treatment in addition to radiotherapy); Recurrence after baseline: recurrence documented from after baseline to 1 month after the second follow-up (No/Yes); Eating ability: scored on a scale of 0–10; CPH: current perceived health, scored on a scale of 1–4; Family income: ≤ 20,000 $HK vs. > 20,000 per month ($1US = $7.8 HK); Occupation: working/not working.
Sample size was derived from 187 participants completing all assessments plus mean substitutions for 24 incomplete assessments.
Treatment (baseline/FU 1)
Recurrence after baseline
Treatment between baseline and FU 1 (β = -0.17, P < 0.01) and recurrence after baseline (β = -0.16, P < 0.01) were found to significantly predict FACT-G (Ch) at FU 2. NPC patients receiving radiotherapy alone reported a better QoL at FU 2 compared with patients receiving combined treatments. Those patients who developed a disease recurrence after baseline were found to have worse QoL at FU 2. Eating ability at FU 1 appeared to predict FACT-G (Ch) at FU 2 marginally. The association between better eating ability approximately 1 month after the completion of radiotherapy and better QoL 4 months later approached but did not achieve statistical significance. Optimism (β = 0.14, P < 0.05) and Worry (β = −0.15, P < 0.05) at FU 1 significantly predicted FACT-G (Ch) at FU 2. NPC patients who scored higher on Optimism at FU 1 had better QoL at FU 2. In contrast, those patients who reported higher levels of Worry at FU 1 were found to have worse QoL at FU 2.
Anger, CPH, Family income, Occupation, and the PSQ-9 did not appear to predict significantly FACT-G (Ch) scores (Table 3). Post hoc analysis of the subscale component of the PSQ-9 that addressed general satisfaction with physicians also was not found to be predictive of the FACT-G (Ch) (β = 0.03, P = not significant). The β value of the five-item MISS was the highest, indicating that it was associated most strongly with FACT-G (Ch) scores.
To exclude contamination of the results by baseline QoL scores, the model was adjusted with the addition of baseline FACT-G (Ch). The most significant predictor after this adjustment remained the MISS at FU 1 (β = 0.20, P < 0.01). Worry, Treatment (baseline/FU 1), and recurrence after baseline also remained significant at the 5% level but Optimism became insignificant (β = 0.12, P = not significant).
To exclude the theory that higher MISS scores reflected a better prognosis, the model was adjusted further for stage of disease at the time of diagnosis. Again, the MISS remained the most significant predictor of prognosis (β = 0.20, P < 0.01). Although Worry, Treatment (baseline/FU 1), and recurrence after baseline remained significant at the 5% level, all other variables were found to be insignificant.
Tolerance levels for all variables in the regression models were > 0.7, with condition indices in all collinearity diagnostics of < 30, reflecting a low possibility of multicollinearity. Thus, the predictor variables appeared to be independent and not likely to be subject to interaction effects. Consequently, no interaction terms were necessary. To test the final regression model fit, residual plots (derived using the Student t test) against the predicted values of the FACT-G (Ch) did not appear to produce any apparent patterns, reflecting a high goodness of fit.
Post hoc analyses exploring disease stage and the MISS revealed no significant difference in mean MISS scores at FU 1 between participants with different stages of disease at diagnosis (F4, 194 =1.0, P = not significant), nor was there any apparent significant difference in the mean MISS scores at FU 1 between the patients who died and those who survived within 6 months after the last follow-up (Student t test = -0.7, df=26; P = not significant).
To our knowledge, the current study is the first to assess longitudinally the effect of satisfaction with the information provided on postradiotherapy QoL in NPC patients. Satisfaction with the information provided reported shortly after the end of radiotherapy was found to be a significant predictor of QoL in NPC patients 4 months later. This finding is consistent with those of a previous study of patients with local NPC12 and studies of patients with breast carcinoma and carcinoma of the head and neck.19, 20 The findings of the current study reflect the benefits of informational support for patients from their physicians as treatment ends, when patients are reassured that they will continue to be monitored and cared for, although clinical consultations will be less frequent.14
Some limitations of the current study warrant discussion. The levels of Anger and Worry changed, as expected, across time. Treatment side effects may induce anger;3 therefore, more intense anger was expected in the postradiotherapy period. The dispositional nature of optimism24 is reflected in the observed stability of scores across time, adding credence to the validity of the measure used. These results are consistent with their expected changes, which helps to validate the single-item measures adopted in the current study. Although multiple measures are preferable, reduced assessment load can be achieved if needed using these single items. The failure to assess formally the performance status of patients with a valid scale precluded a more comprehensive assessment of the patients' functional status.
The drop in family income from baseline to FU 1 indicates that some participants ceased working, if only during radiotherapy. However, by FU 2, there were 20% fewer participants working full time compared with baseline. In the absence of state or private welfare support, the financial costs of being off work dictate the need to resume work as soon as possible after radiotherapy. The QoL level may reflect readiness to resume work.
The five-item MISS appeared to be the best predictor of FACT-G (Ch) scores, being notably better than any medical variables. The importance of psychosocial factors has been documented in long-term rehabilitation studies in patients with carcinoma of the head and neck,34, 35 in which medical variables also were included, although satisfaction with the medical information provided was not measured specifically.
The results of the current study appear to demonstrate that NPC patients may benefit more from information provided by physicians during rehabilitation than was assumed previously. Physician-patient communication can be improved more readily than clinical or occupational factors. Methodologically, the adequacy of the medical measures used in the current study may be a point to consider. It is reasonable to assume that the FACT-G (Ch), which includes numerous psychosocial items, will be found to correlate more with other psychosocial variables such as the five-item MISS than with the medical variables. Therefore, the relative contribution of medical variables in the current study may be underestimated.4
In the current study, tumor stage, an indicator of disease severity, was not found to be a significant predictor of postradiotherapy QoL, a finding that is consistent with those of other reports.4 The likely explanation for this finding is that, regardless of stage, NPC patients are reassured by disease remission after radiotherapy; therefore, disease stage and postradiotherapy QoL may appear to be independent of one another. Optimism has been linked to QoL-related measures22–24 and our measure of Optimism was found to predict postradiotherapy QoL significantly in the current study. The fact that it became insignificant after controlling for baseline QoL and disease stage at the time of diagnosis possibly reflects the impact of prior life satisfaction and disease severity on subsequent optimism toward life. However, it is possible that this item was measuring a proxy of QoL. The six-item physician subscale of the PSQ-9 measures general satisfaction but, unlike the five-item MISS, was not found to be a predictor of QoL, a finding that suggests that the actual information provided and not just general satisfaction with the consultation is important.
The exact reason for this result is unclear. The post hoc findings did not suggest that tumor stage or patient prognosis influenced the relation between MISS and QoL, meaning that good news/bad news differences do not appear to account for the prediction. Approximately 95% of the patients achieved disease remission and the information provided after treatment was nearly certainly positive and encouraging; physicians would be eager to reassure patients all was well. Nearly all patients would have received positive news after treatment. The MISS is unlikely to detect patient satisfaction with the “contact” alone rather than with the amount and style of received information. This finding is based on the fact that the PSQ-9, which also assesses satisfaction with service, was not found to be predictive of QoL, nor was there any substantive association between the MISS and FACT-G “access to doctor” item, which also suggests that access to physicians is not the issue. However, we know information alone may generate better QoL by reducing the patient's uncertainty or anxiety. Therefore it appears that either satisfaction with the information provided, general understanding, or information that lowers uncertainty is responsible for this effect.
To our knowledge, psychosocial research regarding NPC patients is limited. The current study explored the direct relation between satisfaction with physician consultation and QoL in patients with NPC. We believe this is a first step toward a fuller understanding in future studies of the interaction between consultation satisfaction and other potential variables on QoL, given that more robust measures are adopted. We believe that the current prospective study provides further support for the contention that providing information is associated with a better QoL in patients with NPC. The results of the current study reinforce the need to improve physician skills in providing information during consultations with Chinese NPC patients shortly after the end of treatment and suggest that there may be a need for continuing support and information after treatment has ended. The source for this support and information might not need to be a professional. Self-help groups or patient support groups may be an adequate alternative, although it is likely that a professional opinion would be preferred in this population. Further research should explore these preferences.
The authors would like to thank Dr. Fung Hong of HAHO for calling the initial meetings on the project, and Joseph Lau, Ph.D., Joyce Ma, Ph.D., Samuel Ho, Helen Ma, Wong Tai Wai, M.B.,B.S., and Audrey Ho, Ph.D., who were involved in the initial stages of the project. The investigators appreciate the help provided by Dr. Vincent K. C. Tse, and the chiefs of service in the Departments of Clinical Oncology in all five regional hospitals. These included Peter H. K. Choi, W. H. Lau, Damon T. K. Choy, and Anne W. M. Lee. The authors also acknowledge the assistance of Dr. S. F. Leung (PWH), Dr. K. H. Wong (QEH), Dr. Y. C. Foo (QMH), Dr. Candace Ho (TMH), and Dr. Rebecca Yeung (PYH). The investigators also would like to acknowledge the efforts of the interviewers.