Does quality of life among breast cancer survivors one year after diagnosis differ depending on urban and non-urban residence? A comparative study

The Pulling Through Study (PTS) was a longitudinal, population-based study among breast cancer survivors living within 100 kilometers (i.e., 62 miles) of the capital city of Brisbane in Queensland, Australia, and diagnosed in 2002 [20, 21]. This study was extended to include survivors from non-urban areas of Queensland, diagnosed between April 2006 and March 2007 [22]. The Accessibility/Remoteness Index of Australia (ARIA+) classification system was used to define place of residence as either major city, inner regional, outer regional, remote or very remote, and is based on road distance and population size of the nearest town [23]. The selected localities within the perimeter of Brisbane fall within the ARIA+ classification for major cities and hereafter are referred to as 'urban'. Residents of inner regional, remote and very remote areas were pooled as the 'non-urban' group and reflect the reduced access to a range of oncology services experienced by those who live outside state capital cities, irrespective of the level of remoteness [12].

Eligible women, diagnosed with unilateral breast cancer at age 74 years or younger, were randomly selected through the Queensland Cancer Registry (target sample). All cancer diagnoses in Queensland are required to be reported to the Registry and therefore these records provide an accurate sampling frame for recruitment. Since breast cancer is mostly a disease of women 50 yrs or older and to ensure adequate numbers were available for specific age group analyses, younger women were over-sampled in the urban arm of the study, while 100% of eligible non-urban women were recruited for all age groups. Following appropriate ethical approval and the requirements of the cancer registry, doctor consent to contact eligible women (provided for 82% of the urban sample and 90% of the non-urban sample) and participant consent was sought. Overall, 277 urban and 323 non-urban women returned completed quuestionnaires at 12 months post-diagnosis (66% and 71% of eligible women with doctor consent for the urban and non-urban arms, respectively).

Following ethical approval, the general Queensland population sample was derived from the Queensland Cancer Risk Study (QCRS), a population-based survey conducted in 2004 among English-speaking residents of Queensland, aged 20-75 years, randomly sampled within strata defined by gender, age and geographic region (defined by the ARIA+ classification as as major city, inner regional, outer regional or remote/very remote). Further details about the study methods are described elsewhere [24]. Briefly, of the 8,398 adults who agreed to participate in the self-administered questionnaire, 5822 (69.3%) returned surveys, of which 2727 contained QOL information. Analyses reported in this paper include women for whom QOL data were available and who had no prior history of breast cancer, with 675 living in urban and 465 in non-urban areas of Queensland, as defined by the ARIA+.

QOL was measured among women with breast cancer at 12 months post-diagnosis using the Functional Assessment of Cancer Therapy (FACT-G) questionnaire, which is comprised of 27 items rated on a five-point Likert scale (ranging from 0 = 'not at all' to 4 = 'very much') and includes four subscales (physical, social, emotional, and functional well-being). Higher scores represent better well-being. Women in the QCRS received the general population FACT instrument (FACT-GP), which is identical to the FACT-G except it excludes six illness-related items inappropriate for the general population [25, 26]. Overall FACT-GP summary scores and subscales were pro-rated as per the FACT manual to obtain scores comparable to the FACT-G [27], resulting in total scores for all study groups ranging from 0-108 for overall QOL, 0-28 for the physical, social, and functional well-being subscales, and 0-24 for the emotional well-being subscale. Women with breast cancer also completed 13 questions on breast cancer concerns and arm morbidity (FACT-B+4), with total scores from 0-52 for the breast cancer concerns subscale, and 0-160 for overall FACT-B+4. The FACT instrument has excellent reliability and validity [28].

Demographic (age, marital status, educational level, private health insurance, occupation [29, 30] and income), general health (smoking status, body mass index, co-morbidities, complications following surgery, upper-body function [31], physical activity and stress levels including perceived handling of stress) and treatment (chemotherapy, radiotherapy, hormone therapy) characteristics for the breast cancer study participants were also obtained via the questionnaire, whereas information on tumor characteristics were abstracted from histopathology reports (e.g., type of surgery, maximum tumor size and grade, and lymph node status).

Distributions of the FACT scores were approximately normal and hence were summarized as means with 95% confidence intervals (CIs) using SPSS (SPSS Inc, Chicago, IL, version 14). Analysis of variance tests compared age-adjusted mean QOL scores at 12 months post-diagnosis between urban and non-urban breast cancer survivors. Comparisons between breast cancer survivors and women from the general population involved general linear regression models to obtain QOL scores adjusted for characteristics that differed between the groups (i.e., potential confounding factors).

Descriptive results presented in this study have been adjusted for the sampling fraction used to identify younger breast cancer patients from urban areas (weighting applied: < 50 years:1.0; ≥50 years:1.3). The general population comparison group was also weighted by age, based on Australian Bureau of Statistics data, so that results reflect the actual female Queensland resident population (weighting applied for urban, regional, outer regional, remote and very remote: < 50 years:1.3, 1.3, 1.4, 1.5 and 0.9, respectively; ≥50 years:0.8, 0.8, 0.7, 0.6 and 1.1, respectively) [33]. The conventional P < 0.05 level (two-tailed) was accepted as statistically significant. Differences of eight or more points in mean FACT-B+4 scores, five or more points in mean FACT-G scores, three or more points on the breast cancer concerns subscale and two or more points for all other subscales between urban and non-urban breast cancer survivors or between women with breast cancer and their general population peers were considered clinically important, as recommended by developers of the FACT [25]. For correlates, a difference in odds ratios (ORs) of ≥1.8 or ≤0.6 was considered to be of potential clinical relevance.

As suggested by Fayers [34], a new outcome measure was calculated to characterize breast cancer survivors whose QOL was below normative levels. QOL values were calculated for each five-year age stratum of the general population study group and subtracted from the QOL score within the same age group of women with breast cancer (i.e., case FACT-G minus general population comparison group FACT-G) separately by urban and non-urban residence [34]. Positive scores indicate higher QOL, and negative scores indicate lower QOL, among cases relative to age- (within five years) and residence-matched peers. Relative overall QOL (FACT-G) was then categorized into groups using score differentials considered clinically important to investigate the proportions of breast cancer survivors with relative overall QOL lower than (-5.0 points or more), similar to (>-5.0 to < +5.0) or better than (+5.0 points or more) the general population study group. Relative QOL was also calculated for each subscale, using two points as the critical threshold. A dichotomous outcome variable was defined, combining the 'similar' and 'better' groups, and binary logistic regression was used to generate ORs and 95% CIs to identify demographic, general health, and clinical characteristics associated with QOL status below the norm compared to the 'similar/better' group. A range of potentially important correlates were explored, however, only those that were found to be statistically significant or clinically important are reported. Formal tests of interactions between residence and each of the characteristics of interest did not yield any statistically significant results, therefore pooled results, adjusted for residence, are presented.

Demographic and disease characteristics were similar for the women with breast cancer in this study and those in the target sample. The majority of women (75-80%) were diagnosed with infiltrating ductal carcinoma, approximately 60% received complete local excision of their tumour and more than 50% had 10 or more lymph nodes removed. However, participants among urban breast cancer survivors had somewhat smaller tumor size (median tumour size was 14 mm) when compared with the target sample [20‐22]. For the majority of demographic and general health characteristics, women with breast cancer had similar characteristics, irrespective of place of residence. However, the urban compared to the non-urban breast cancer sample was more likely to be unmarried, have private health insurance and report fewer co-morbidities, and less likely to be obese (Table 1). Non-urban compared to urban women with breast cancer were more likely to have multiple forms of adjuvant therapy and less likely to report multiple complications.

A comparison of women with or without breast cancer showed significant differences for several demographic and general health characteristics (Table 1). Breast cancer survivors tended to be older or have lower educational levels when compared with the general population, irrespective of residence. In addition, urban breast cancer survivors were more likely to be single, have private health insurance, and/or fewer co-morbidities (other than breast cancer), while non-urban breast cancer survivors were more likely to be sedentary and/or have two or more co-morbidities (other than breast cancer), when compared with their general population counterparts. While there was a significant (P < 0.05) difference in body mass index between urban breast cancer survivors and their general population peers, this was attenuated when missing values were omitted from analyses.

Although urban breast cancer survivors reported higher age-adjusted QOL summary and subscale scores than their non-urban counterparts 12 months following diagnosis, differences did not reach the threshold for clinical importance even for those subscales that were statistically significant (physical, emotional, and overall QOL, P < 0.01). In contrast, well-being related to breast cancer concerns was lower among non-urban compared to urban survivors by statistical (P < 0.01) and clinical criteria (Table 2). Furthermore, for both groups, women reported most detriment to their QOL for this subscale, with participants reporting mean values below 80% of the maximum score on average. Participants reported mean values at approximately 80% of the maximum score for all other subscales.

Table 3 presents the subscale and overall mean FACT-G scores for breast cancer survivors compared with women from the general population, stratified by residence and adjusted for potential confounding factors. At 12 months post-diagnosis, urban and non-urban breast cancer survivors reported clinically higher social well-being compared with their general population peers, and non-urban breast cancer survivors also reported clinically lower physical well-being (P < 0.01 for all). Scores for emotional, functional and overall (FACT-G) QOL were clinically comparable to their counterparts from the general population despite a statistically significant difference for emotional and functional well-being (P < 0.01).

Using the new outcome measure of QOL relative to age and residency-matched women from the general population, depending on the specific QOL scale, between 17.2% and 32.8% of all women with breast cancer reported clinically lower QOL 12 months following diagnosis than age- (within five years) and residence-matched women without the disease. A further 17.5%-48.5% of women reported similar QOL, while the remainder (19.8%-65.3%) reported clinically better QOL (Figure 1). The subscales with the highest proportions below the norm were emotional (32.8%) and physical (29.3%) well-being, and overall QOL (26.2%).

Following adjustment for potential confounding factors, a range of characteristics were associated with breast cancer survivors reporting overall (FACT-G) QOL below normative levels (Table 4), but place of residence (i.e., urban versus non-urban) was not one of these (Odds Ratio (OR) = 1.06; 95% Confidence interval (CI) = 0.64-1.74). Experiencing one or more complications following surgery was associated with two-fold increased odds (OR = 2.26, 95% CI = 1.31-3.90; P < 0.01) of reporting reduced QOL, while upper-body function below the median, moderate or higher stress levels and poor perceived handling of stress were each associated with at least four-fold increased odds of reporting reduced QOL (ORs ranging from 4.24-4.77, P < 0.01, see Table 4). A marker of higher socioeconomic status, having private health insurance, was associated with a 0.6 odds of reporting lower relative QOL (95% CI = 0.37-0.99, P = 0.05).