How Does Sex Influence Multimorbidity? Secondary Analysis of a Large Nationally Representative Dataset

Multimorbidity increases with age and is generally more common in women, but little is known about sex effects on the “typology” of multimorbidity. We have characterized multimorbidity in a large nationally representative primary care dataset in terms of sex in ten year age groups from 25 years to 75 years and over, in a cross-sectional analysis of multimorbidity type (physical-only, mental-only, mixed physical and mental; and commonest conditions) for 1,272,685 adults in Scotland. Our results show that women had more multimorbidity overall in every age group, which was most pronounced in the 45–54 years age group (women 26.5% vs. men 19.6%; difference 6.9 (95% CI 6.5 to 7.2). From the age of 45, physical-only multimorbidity was consistently more common in men, and physical-mental multimorbidity more common in women. The biggest difference in physical-mental multimorbidity was found in the 75 years and over group (women 30.9% vs. men 21.2%; difference 9.7 (95% CI 9.1 to 10.2). The commonest condition in women was depression until the age of 55 years, thereafter hypertension. In men, drugs misuse had the highest prevalence in those aged 25–34 years, depression for those aged 35–44 years, and hypertension for 45 years and over. Depression, pain, irritable bowel syndrome and thyroid disorders were more common in women than men across all age groups. We conclude that the higher overall prevalence of multimorbidity in women is mainly due to more mixed physical and mental health problems. The marked difference between the sexes over 75 years especially warrants further investigation.


Introduction
Multimorbidity is usually defined as the co-occurrence of two or more long term conditions within one individual, and is an increasing problem with a reported doubling of prevalence over the last 20 years [1]. It is a global phenomenon and has now become the norm rather than exception in many populations [2][3][4][5]. Multimorbidity is associated with increased mortality [6][7][8], lower quality of life [9][10][11], and greater utilization of healthcare including unplanned admissions [12][13][14] and thus higher healthcare costs [6,[15][16][17]. Patients with multimorbidity are less satisfied with care provided [18], perhaps due to the fragmentation of care resulting from the single disease-focus that drives much of current health care [19][20][21].
The prevalence of multimorbidity rises rapidly with increasing age [4][5][6]16,[22][23][24], although, the absolute number of individuals with multimorbidity is often higher in those under 65 years [4,25,26]. It is also strongly associated with socioeconomic deprivation and occurs 10-15 years earlier in individuals living in the most deprived compared with the least deprived areas [3][4][5]. The effect of sex on multimorbidity has been less well defined [3]. Most previous studies have shown an increased prevalence of multimorbidity among women [1,6,16,26], though not all studies find this [2,27,28]. Although in almost all countries in the world women have a longer life expectancy than men [29], they are more often affected by a number of non-fatal chronic diseases that decrease quality of life and everyday physical ability [23,30].
Multimorbidity is a broad concept, and can be characterised further in a number of ways, such as physical-only, mental-only, and mixed mental and physical [24], as well as by the commonest combinations of individual conditions. The typology of multimorbidity by sex has not been well documented in large primary care populations. Therefore, this study aimed to examine the relationship between sex and different types of multimorbidity across different age groups, using cross-sectional data from a nationally representative large primary care dataset.

Methods
We obtained a dataset for 1,751,841 patients of all ages, who were alive on 31 March 2007 and permanently registered with 314 Scottish general practices, from the Primary Care Clinical Informatics Unit at the University of Aberdeen, UK [31]. This accounts for about a third of the Scottish population and is representative of it in terms of age, sex, and socioeconomic deprivation. The NHS National Research Ethics Service had previously approved the use of these anonymised data for research purposes and this analysis did not require independent review [4].
Forty conditions in total were analysed as previously defined [4] using either Quality and Outcomes Framework (QOF) register Read Code sets where available (Read Codes are the clinical coding system used in UK general practice) [32], other Read Code sets used by NHS Scotland for non-QOF conditions, or combinations of Read Codes and prescription data where coding was known to be likely to under-record important conditions. The appendix provides further detail of definitions and the 40 conditions analysed.
Multimorbidity was defined as the presence of two or more disorders in an individual. The choice of conditions was based on recommendations from systematic review for multimorbidity measure [33] as well as diseases in QOF within UK General Practice (GP) contract and important diseases identified by NHS Scotland [32,34]. To specifically examine multimorbidity in terms of physical and mental health disorders, we defined each condition as either a physical or mental health disorder. Of the 40 conditions we classified 32 as physical and 8 as mental disorders (depression, anxiety, drugs misuse, alcohol misuse, schizophrenia/bipolar, dementia, learning disability and anorexia).
In order to characterize the prevalence and type of multimorbidity, we additionally analysed how many people had: (a) 2 or more physical conditions but no mental health conditions (physical-only); (b) 2 or more mental health conditions but no physical conditions (mental-only); or (c) 2 or more conditions including at least 1 physical and 1 mental (mixed physical and mental). We then examined the top 10 most common conditions in those with multimorbidity separately in each age group by sex.
As the current analysis focused on adults with multimorbidity, we excluded adults aged under 25 years in whom multimorbidity is uncommon (prevalence 1.9%) [4]. We divided the remaining patients into 6 age groups (25-34, 35-44, 45-54, 55-64, 65-74 and 75 and over) to reflect different stages in the life course. For each of these age groups, we calculated the number (and percentage) of patients with: no conditions; with one condition; and with two or more conditions. We used graphical display to compare differences by age group in patterns of multimorbidity by sex. We calculated the ten most prevalent conditions by age group and separately for men and women with two or more conditions by dividing the number of patients with each condition by the number of patients in the relevant age groups with two or more conditions. All analysis was conducted using Stata version 13.0 (Stata, College Station, TX, USA).

Results
The analysis involved 1,272,685 adults aged 25 years and over, of whom 51% were women and 49% were men. Table 1 shows differences in prevalence of morbidity by age group and sex. For both men and women, the number and percentage of people with multimorbidity was higher in each successive ten-year age group within each sex, however, there were more people with multimorbidity in absolute numbers under the age of 65 (201,311 vs. 194,996). There were more women with multimorbidity than with a single condition or no condition in all age groups above the age of 55, whereas this was only true in men above the age of 65. The prevalence of multimorbidity (two or more conditions) was higher in women than in men in every age group, although the differences were small over the age of 65. The biggest sex difference was found in the 45-54 age group (women 26.5% vs. men 19.6%; difference 6.9 (95% CI 6.5 to 7.2). Table 2 shows the prevalence of different types of multimorbidity (physical or mental or mixed) across age groups by sex. In general, the differences in prevalence of all types of multimorbidity between men and women were markedly small. However, whilst physical-only morbidity increased with age for both men and women, differences were small up to the age of 64 years but were marginally higher for women. After the age of 65 more men than women had physical-only multimorbidity. The most significant difference was observed in the over 75 age group (women 45.3% vs. men 53.5%; difference´8.3 (95% CI´7.7 to´8.7). Gender differences were greatest for mixed physical and mental multimorbidity (as compared with mental-only and physical-only) at all ages but the magnitude of this difference varied with age. Women's higher prevalence of mixed physical and mental multimorbidity was small in absolute terms at age 25-34 (2.1% (95% CI 1.9 to 2.2), increasing to around 5% between the ages 45 and 74), and highest in those above the age of 75 (women 30.9% vs. men 21.2%; difference 9.7 (95% CI 9.1 to 10.2). Figure 1 shows the number of people with the different types of multimorbidity by age group and sex. It illustrates that, overall, women have more multimorbidity across the life course but sex differences are small and diminish even more after 55-64 years. It shows that a large part of this reduction in sex difference is due to a steeper rise in the prevalence of men with physical only multimorbidity compared to women. Table 3 shows prevalence (%) of the conditions which feature in the top 10 most commonly found in men or women with multimorbidity within each age group. For women with multimorbidity below the age of 55, depression was the condition with the highest prevalence, whereas above the age of 55 hypertension was the most common condition. In contrast for men with multimorbidity, drugs misuse had the highest prevalence for men aged 25-34, depression for men aged 35-44, and hypertension for men aged 45 and over. Women had a consistently higher prevalence of depression, pain, IBS and thyroid disorders across all age groups. For men under 45 with multimorbidity, compared to women in the same age group, prevalence was only notably higher for substance (alcohol and drugs) misuse. Amongst people with multimorbidity, higher prevalence's of coronary heart disease (CHD) and diabetes were recorded in men than women from the ages of 45 and over.

Discussion
This study has examined patterns of multimorbidity by age and sex in 1,272,685 adults in a nationally representative primary care sample in Scotland. Women had more overall multimorbidity than men at all ages, although the sex differences were relatively small. From the age of 45, physical-only multimorbidity was consistently more common in men, and physical-mental multimorbidity more common in women. The biggest sex difference in physical-mental multimorbidity was found in the 75 years and over group. The commonest condition in women was

Discussion
This study has examined patterns of multimorbidity by age and sex in 1,272,685 adults in a nationally representative primary care sample in Scotland. Women had more overall multimorbidity than men at all ages, although the sex differences were relatively small. From the age of 45, physical-only multimorbidity was consistently more common in men, and physical-mental multimorbidity more common in women. The biggest sex difference in physical-mental multimorbidity was found in the 75 years and over group. The commonest condition in women was depression until the age of 55 years, thereafter hypertension. Depression, pain, irritable bowel syndrome and thyroid disorders were more common in women than men across all age groups.

Relationship with Existing Literature
Several studies have reported that women have more multimorbidity than men [4,26,35], as reflected in a recent systematic review [3]. The level of sex difference in multimorbidity varies between studies. One study reported little difference between the sexes on the risk of being mutlimorbid [36]. This study used 20 chronic conditions and the difference in reported results may be explained by the inclusion in our study of additional conditions which are far more prevalent in women such as migraine and thyroid diseases. Moreover, most previous studies have had either much smaller samples sizes [26,27,36,37] or were restricted to elderly [38][39][40] or hospitalised people [41], whereas the current study examined a nationally representative one-third of the Scottish population.
As far as we are aware, our study is the first to show prevalence rates of different types of multimorbidity and particular conditions by sex and across all ages in a large, nationally representative dataset.
In terms of mental health morbidity, our results show that physical-mental multimorbidity is much more common in women than men. Depression was more prevalent in women than men and was the most frequently occurring condition among women below the age of 55. These findings agree with previously published studies in both multimorbid [4,42] and single disease analyses. [43] Whether this reflects "true" levels of depression in patients with at least one other condition, or a difference in the presentation (by patients) or recognition (by patients and doctors) of depression in people with other long-term conditions cannot be determined from these data. Experimental studies support differential recognition by doctors of potential symptoms in women and men [44][45][46], at least for cardiovascular disease and depression. In addition, below the ages of around 60 years, women are more likely to have contact with their general practitioner [47], and there thus may be differences between men and women of prior recording of depression which may alert GPs to be more alert to depression when they see female patients with physical conditions. Recent survey data have shown that 90% of women with depression discussed it with their doctor [48]. Underestimation of depression in men could perhaps also be explained by so called "masculine" ways of responding to negative affect, less help seeking and lack of consideration for some of such symptoms in commonly used diagnostic tools for depression [49]. This may be important particularly in older populations as previous literature has described that depression is related to negative outcomes within chronic diseases such as, e.g., coronary heart disease [50].
We also found higher rates of alcohol misuse for men with multimorbidity, particularly in the younger age groups, which have also been reported elsewhere [51,52]. This is perhaps not surprising given the consistent finding of higher consumption of alcohol in general population samples of men compared with women in many cultures [53]. It is also consistent with the view that men's mental distress and depressive illness may disproportionately present as substance abuse [54,55].
Although we did not examine the utilization of healthcare in the current study, increased multimorbidity among female patients, especially those with mental health problems, could explain the previous results in literature in relation to consultation rates and emergency inpatient admissions [12,47]. A recent study using a large dataset in England reported that consultation rate among women was higher than men, which was only partially explained by reproduction related visits [47].

Strengths and Limitations
As far as we are aware, the present study represents the most comprehensive study to date on sex associations with multimorbidity, based as it is on a large nationally representative primary care sample. The study has, however, several limitations. Firstly, it was based on electronic medical records from GP practices and thus relies on the diligence of the practitioners in coding conditions, and their propensity to do this equally diligently by sex for all conditions. We used mixtures of morbidity codes and prescribing to define and measure some conditions, firstly to ensure that coded conditions were still "active" (e.g., asthma code plus recent inhaler prescription) and secondly because some conditions not incentivised in the UK GPs Quality and Outcomes Framework (such as osteoarthritis) are poorly recorded. In addition, the physical conditions we used did not include chronic conditions that are specific to women (in order to compare men and women across conditions). The study is also not able to take account of severity of any of the conditions as these data were not available. Additionally, we did not examine ethnic differences in multimorbidity.

Implications for Policy and Practice
This study shows that multimorbidity differs between men and women in terms of types of conditions as well as prevalence, raising the potential need for a more tailored gendered approach to patients with multiple chronic conditions. Lack of appropriate guidelines and integrated health systems creates a gap between patients' healthcare needs and experiences and its provision [55]. As reflected in previous studies, this gap could be decreased by improved integration and a generalist patient-centred approach [56]. Given the apparently large additional multimorbidity burden in women relating to sex-specific conditions, Gender sensitive services development may be important in some settings, including the implementation of sex related health and socio-economic risk factors reduction policies as suggested by WHO [57,58] Further research is required to understand the differing burdens and impacts of multimorbidity typologies in men and women.

Conclusions
Our study has shown that, overall, women have more multimorbidity across all age groups, which is driven by a higher prevalence of combined physical and mental multimorbidity in women than men, with the biggest difference in this category being found among the oldest (above the age of 75 year). This warrants further study, including the severity, burden and effects of such sex differences in multimorbidity.