Gender differences in health-related quality of life of Australian chronically-ill adults: patient and physician characteristics do matter

In 2004, 77% of Australians reported having at least one long-term medical condition [1]. The management of chronic illness has become a major focus in general practice in Australia, both because of its prevalence and because of the opportunity for general practice to intervene early to improve quality of life, prevent disability, and reduce hospital use [2].

The SF-36 and SF-12 are widely used measures of health-related quality of life (HRQoL). Investigators from numerous countries representing diverse cultures have determined that both measures are sensitive to differences in a number of socio-demographic and clinical variables, including gender [3, 4], age [2, 3], income [4‐6], employment [2, 4, 5], education [5, 6], marital status [2, 7], satisfaction with care [2], smoking [8‐10], and number of chronic conditions [6]. Several studies have found women to have lower HRQoL scores than men [11, 12] and a few studies have found that smoking had a greater negative effect on HRQoL of women than men [13]. Employment had a stronger association with the quality of life of males than that of females [2]. In addition to patient factors, we also included the socio-demographic characteristics of the general practitioners (GPs). Earlier research has shown that female GPs provided longer consultations and saw patients with different health problems than their male colleagues [14, 15]. Gender is one of the many factors that impact upon the doctor–patient interaction [16]. From the above findings, patient and GP characteristics tended to interact with gender of patients in predicting HRQoL, so we decided to analyse data separately for males and females.

In this study, we investigated the relationship between patient or GP characteristics and HRQoL in a large sample of chronically-ill Australian adults from two states and the Australian Capital Territory, using SF-12 version 2. No one we know of has investigated the impact of characteristics of general practitioners on their patients’ HRQoL before. Based on the findings and identified gaps in the literature, the following research questions were posed:

The mean ages of male and female patients were 64 years (range 24–90) and 63 years (range 18–93) respectively.

The gender comparison for socio-demographic characteristics of patients and GPs is presented in Table 1. Female patients tended to have lower socio-economic status (13% more unemployed, 10% less owner-occupiers with cars, 6% more lower-educated) than male patients (Table 1; P < 0.001). Married patients were more likely to be males (78% vs 64%, P < 0.001). Male patients were more likely than females to have more than one chronic disease (P < 0.001). Female GPs were more likely to have female than male patients (33% vs. 15%, P < 0.001) and the proportions were 37% vs. 18% (P < 0.001) without solo practices. Female patients tended to be under the care of younger (practicing for < 19 years) GPs than male patients (37% vs 29%, P < 0.001).

The overall means of PCS-12 scores for male and female patients were 43.0 (SD = 11.7) and 42.3 (SD = 12.4) respectively (Table 2). Similarly, overall MCS-12 scores were 50.0 (SD = 10.9) for men and 48.3 (SD = 11.2) for women. Table 2 shows the differences between unadjusted PCS-12 or MCS-12 scores of male and female patients for the subcategories of patient and GP characteristics (Table 2). Women reported poorer physical health than men among those who had one chronic disease (P = 0.035) and were under the care of a male GP (P = 0.014, Table 2). Unemployed females reported better physical health than unemployed males (P = 0.020). Similarly, women reported poorer mental health than men across the following patient characteristics: middle age (P = 0.005), home and car owners (P = 0.007), employed (P = 0.003), two or more chronic diseases (P = 0.01), smokers (P = 0.001), high or low PACIC scores (P = 0.03), under the care of a male GP (P = 0.017), a GP who was Australian trained (P = 0.001), < 6 (P = 0.028) or > 15 years (P = 0.003) as a GP in the current practice, > 24 years as a GP (P = 0.013), who worked in a larger practice (P = 0.006) or in an urban area (P < 0.001) (Table 2). However, the above effects were not adjusted for confounding effects.

Gender differences in independent variables after adjustment for confounding effects with the multilevel regression analyses for each of the response variables are presented in Tables 3 and 4.

PCS-12 scores declined with age for both male (P < 0.05) and female patients (P < 0.01), but in contrast MCS-12 scores increased with age for female patients (P < 0.001).

Both men and women with good general health were likely to have clinically significant better physical health with large effect sizes (> 1.3, B (regression coefficient) > 10.9, P < 0.001) and better mental health with a moderate effect size (> 0.7, B > 6.8, P < 0.001) compared to those with poor general health (Tables 3 & 4). Home and car ownership was positively associated with PCS-12 scores of female patients (effect size = 0.79, B = 4.2, P < 0.01) and MCS-12 scores of male patients (effect size = 0.82, B = 3.4, P < 0.05). Patients who were employed (effect size = 1.23, B = 7.3, P < 0.001) or retired (effect size = 0.51, B = 3.9, P < 0.001) were likely to have higher PCS-12 scores and MCS-12 scores (effect size > 0.55, B > 3.4, P < 0.01) than unemployed male patients. Employment was likely to have a positive effect on PCS-12 scores (effect size = 0.78, B = 4.1, P < 0.001) and retirement was likely to have a positive effect on MCS-12 scores (effect size = 0.52, B = 2.9, P < 0.01) for women. An effect size of more than 0.5 is considered to be clinically significant [3]. The number of chronic medical conditions was negatively associated with both MCS-12 (P < 0.05) and PCS-12 scores (P < 0.05) for women and with PCS-12 (P < 0.001) for men. Female patients who were married or cohabiting tended to have higher MCS-12 scores than those who were not (P < 0.001). Female smokers were likely to have a lower mental health score than non-smokers (effect size = 0.66, B = −4.0, P < 0.001). Neither smoking status nor marital status was associated with PCS-12 scores for any gender or MCS-12 scores for men. Results also showed a positive association between patient assessed quality of care and MCS-12 scores (P < 0.05) and a negative association between quality of care and PCS-12 scores (P < 0.05) for females only. Effect sizes for some of the significant effects (for example, number of conditions, satisfaction with care, married status) were small (< 0.5) and clinically not significant. Being under the care of a female GP was likely to have a positive effect on PCS-12 scores (P < 0.05) of female patients. Other GP characteristics were not associated with either PCS-12 or MCS-12 scores for either gender.

This study is one of very few to explore gender differences in HRQoL among patients with type 2 diabetes and hypertension/ischaemic heart disease. The SF-12 is a subjective measure of health that can be influenced by a respondent’s perceptions, expectations and interpretations about health [6]. Nonetheless, the scale has become one of the most widely used HRQoL measures. This study provides comprehensive data on how patient and GP characteristics predict self-rated physical and mental health of chronically-ill patients in Australia.

The results show that chronically-ill women smokers reported clinically significantly (effect size = 0.66) poorer mental health than non-smokers. This supports findings from previous studies showing that the association between smoking and HRQoL is different between women and men [13, 30, 31]. Lasser et al. [10] suggested that people with poor mental health are more likely to smoke than those who have good mental health. A higher physical and psychological dependence among women is a possible explanation for the increased mental distress observed among women who unsuccessfully attempted to quit [32]. Strine et al. [8] found a significant association between smoking and impaired mental health. The provision of psychological care in conjunction with smoking-cessation programs, and vice versa, is indicated [33]. Smoking is one of the strongest modifiable risk factors for a host of health outcomes that contribute to morbidity and mortality in Australia and worldwide in chronically-ill patients. All this suggests that chronically-ill women may need more psychological support in their attempts to quit smoking. Smoking is one of the most significant risk factors for the development of cardiovascular disease in diabetes patients [34]. Smoking is a risk factor for mortality and coronary heart disease in hypertension and in diabetes [35]. Chronically ill, particularly females with mental illness are motivated to attend smoking reduction and cessation programmes. When supported programmes providing nicotine replacement and counseling are offered to people with mental illness, it has been found that they are able to quit smoking at equivalent rates to the general population [36]. It is important to ensure that the person with severe mental illness is aware of the risks of smoking because basic medical education is frequently missing in this patient group [37]. Integrated care is desirable, as psychiatric symptoms may be exacerbated and severe withdrawal symptoms experienced by smokers with a mental illness undertaking smoking cessation treatment. When smoking cessation and psychiatric care is integrated these adverse effects appear to be reduced [38]. The ability to detect hypothesized relationship in previous studies between smoking and mental health of females suggests construct validity of smoking status in this study.

The results show that women under the care of female GPs reported better physical health (effect size = 0.30) than those under the care of male GPs. Physician’s gender is one of many factors that impact on the doctor patient interaction [16]. As noted in previous studies, female GPs tend to have longer consultations [15, 39], especially with their female patients. Female patients often report that they prefer female doctors for female-specific health problems [40, 41], intimate problems [42, 43], behavioral problems [41], and endocrinologic/metabolic problems [14]. Female physicians were found to be more likely to perform female-specific prevention procedures, check patients’ blood pressure and make some follow-up arrangements and referrals [44]. Also, compared to their male colleagues, female physicians were shown to use a more participatory decision-making process [45] that encourages self-management, which is a fundamental part of diabetes care and may result in improved diabetes control [46].

Socio-demographic differences between male and female patients in the current study (e.g., the women were younger, less likely to be married, and had lower socio-economic status than the men) are consistent with those found in other similar studies [47, 48] and provide an understanding of gender differences in HRQoL.

Our study showed that there are gender differences in how patient characteristics impact on self-assessed physical and mental health. For example, previous research found that lower socio-economic groups reported lower PCS-12 and MCS-12 [4, 5, 49]. Our results showed that home and car ownership tended to have a positive effect on the self-assessed physical health of women but not men and on the mental health of men, but not women. Some studies have shown a significant interaction effect between gender and employment, indicating that employed men enjoyed higher levels of general well-being [2, 7, 50]. In this study the negative impact of unemployment was likely to be greater in male than female patients. Unemployment was likely to have a larger negative effect on HRQoL of men than that of women. This may be because the significance of work and its impact on household income may be greater in chronically-ill older men than in women [51]. Also the observed age difference in HRQoL differed among men and women. The younger men (< 39 years) were likely to report better physical health than the older men (> 59 years), whereas the older women reported better mental health than the younger women.This result is consistent with previous studies. For example, Hanmer et al. [52] reported similar mental health (MCS-12) mean scores for younger men (< 39 years, 52.1) and older men (> 59 years, 52.3) whereas older women reported better mental health mean scores (51.3) than younger women (49.6).

Two hundred and eleven patients suffered from only type 2 diabetes and 793 from type 2 diabetes and hypertension and/or ischemic heart disease. Further 1129 suffered from hypertension and/or ischemic heart disease. The overall mean PCS-12 of male diabetes only patients (46.7, SD = 12.4) and male hypertension/ischaemic heart disease only patients (44.8, SD = 11.5) in the study were less than male U.S. general population (mean = 50.6, SD = 10.2) [19]. The difference for PCS-12 was clinically not significant for diabetes (effect size =0.38 < 0.5), whereas it was clinically significant (effect size = 0.56) for hypertension/ischaemic heart disease. Similarly, the overall mean PCS-12 of female diabetes only patients (44.7, SD = 11.6) and female hypertension/ischaemic heart disease patients (43.4, SD = 12.6) were less than female U.S. general population (mean = 48.7, SD = 9.6) [19]. The difference was clinically not significant for diabetes (effect size =0.41), whereas for hypertension/ischaemic heart disease patients, the difference was clinically significant (effect size = 0.53). However, mean MCS-12 of male (49.3, SD = 10.6) and female (47.7, SD = 12.3) diabetes only patients and male (50.5, SD = 10.5) and female (49.2, SD = 10.8) hypertension/ischaemic heart disease patients was not clinically significant compared to male (50.4, SD = 9.9) and female (48.4, SD = 9.6) MCS-12 scores of U.S. general population.

Although the GP level variance explained was very high (from 70% to 90%) for both summary scores, the patient level variance explained for mental health was half that of physical health (36%). There may have been other lifestyle and clinical risk factors important to mental health assessment which were not specifically evaluated in this study and warrant further exploration in the Australian context.

There are a number of limitations to this study. Patients that the practice identified as being unable to read English were excluded from the study. Although the response rate of 70% was comparable with other studies [53], it is possible that non-responders might have assessed their physical and mental health differently from those who responded. We were unable to analyse differences between respondents and non-respondents in the study as information about non-respondents could not be captured as recruitment was at arm’s length through practices. In a similar HRQoL study in Australia with 7606 chronically-ill patients from 96 general practices, the response rate was 61% [2]. In that study we had gender for non-respondents. We conducted analyses comparing proportions of respondents with non-respondents for gender. The gender of respondents (53.3% were females) and non-respondents (53.6% were females) were similar (P = 0.76). As females attend a GP more often than males, they have a greater chance of being selected in the sample [54]. Male (46.7%) and female (53.3%) patients responded to mental health questions in the study were similar to other studies [2]. Further, the socio-economic status of male patients were similar in both studies (home owners: 82.3% vs. 81.4%). We compared patient characteristics of this study with a similar Australian general practice study with type 2 diabetes, ischaemic heart disease/hypertension and asthma patients [55]. The proportions of gender (P = 0.712), home ownership (P = 0.690), and marital status (P = 0.903) were similar between the two studies (data not shown). However, the patients in this study were older (69% vs. 55% for > 59 years, P < 0.01) and marginally less employed (32% vs. 34%, P = 0.024). The reason being that, in the other study asthma patients (n = 724) were much younger (mean = 50 years) and our study does not have any asthma patients. The mean age in our study was 64 years. Hence our study had more older and retired (47% vs. 40%), and therefore less employed patients.

In Australia, the proportion of males and females who smoked declined 1.4% (22.5% in 2004 and 21.1% in 2007) and 1.1% (18.8% in 2004 and 17.7% in 2007) respectively between 2004 and 2007 [56]. In Australia, the proportion of males and females type 2 diabetes and IHD patients who smoked in 2007 were 13.9% and 15.3% respectively [57]. Those figures were slightly higher than the proportions for type 2 diabetes and/or IHD patients for males (9.2%) and females (11.1%) of our study.

There were no significant differences in GP characteristics between our sample and all GPs in Australia in terms of female gender (37.3% vs. 35.9%). This proportion was similar in Beech study (36.8%) [54]. However, GPs who graduated in a country other than Australia were slightly under-represented when compared with Australian total sample (25.5% vs. 30.2%) but not with those participated in Beech study (26.5%) [54]. In Australia, patients can choose what doctor they see, at whichever practice they choose. There are no patient boundaries and patients do not have to ‘sign up’ with a particular practice.

The actual ICC computed for MCS-12 from the final multilevel model is 0.003 for females and 0.009 for males which is lower than that used in the power calculations. With actual ICC, 960 patients from each gender were adequate to detect an effect size of 0.13 between smokers and non-smokers.

Strengths of the study include the large number of patients and GPs participating, the adjustment for confounding patient and GP factors, the correction for GP-level clustering with multilevel modeling, and addressing the potential bias and loss of precision arising from missing values using multiple imputations.

The increased prevalence of poorer self-assessed mental health status among chronically ill women who smoke is deserving of further investigation. Better understanding of the associations between HRQoL and smoking may help to reduce barriers to smoking cessation and help to better tailor smoking cessation support to patients’ needs, ultimately improving their overall well-being. This is especially important in the context of chronic disease, where improvement of lifestyle behavior is central to management. Clinicians are advised to pay more attention to the mental health status of chronically-ill smokers, specially women.

More quantitative studies are needed to explore the interaction between GP gender and patient gender in relation to HRQoL.