Introduction

Management of the increasingly common problem of multimorbid chronic diseases is a major global healthcare challenge [1]. Multimorbidity is assisted with functional decline, an increased risk of mortality and an increased use of healthcare resources [2–5]. Compared to multiple chronic physical diseases, physical and mental health comorbidity is associated with even greater functional decline [6]. Healthcare approaches focused on single disease are unlikely to effectively coordinate healthcare for the complex needs of people with multimorbidity [7–10]. Whilst multimorbidity is known to increase with age and socioeconomic deprivation, estimates of the prevalence of multimorbidity are heterogeneous; most studies have counted small numbers of morbidities, are focused on older people and hospital populations, physical health problems and high-income countries [11–13].

The World Health Organization´s (WHO) global action plan for prevention and control of chronic diseases emphasises low and middle-income countries, where the majority of world population lives, socioeconomic deprivation is more common and where 80% all deaths due to chronic disease occur [14]. Recent, first multi-national investigation in low and middle-income countries reported prevalence rates of multimorbidity in older people that varied widely [13], suggesting that multimorbidity needs better understanding in the context of country-specific characteristic, such as demographic structure, disease burden and health system.

Brazil is a rapidly developing middle-income country with the world’s fifth largest population [15] and the rising incidence of chronic non-communicable diseases [16]. The estimated economic cost due to deaths from chronic non-communicable diseases and productivity losses due to absenteeism and presenteeism in Brazil in 2015 equated to a total of 5.4% (US$129.8 billions) of Gross Domestic Product [17]. To meet the demand for cost effective, efficient, safe, high quality provision of comprehensive care, Brazil has progressively strengthened the Family Health Programme and recently established operation norms for a regionalised network of health services [18, 19]. The coordination of the regionalised network of health services remains complex as it includes distinct but interconnected public and private sectors [19]. A specific action plan for prevention and control of multimorbidity through this highly complex system is currently lacking and its development is hindered partially by limited evidence. In Brazil, multimorbidity occurrence and patterns has been studied only in relatively small samples of the general population of one city [20], in older people [21] and in women [22]. Even less is known about mental health comorbidity specifically in Brazil [23], and this is despite the 12-month prevalence rate of major depressive episode in Brazil being estimated to be the highest in the world [24].

Better understanding of multimorbidity in Brazil can inform interventions to prevent and control multimorbidity, reduce its burden, and align health-care services more closely with patients’ needs in Brazil and in comparable middle-income countries. We aimed to use a large, nationally representative sample of Brazilian adults to explore the distribution of multimorbidity and to identify patterns of multimorbidity of chronic physical and mental health conditions.

Materials and methods

Results

The study population was likely to be aged 25–44, live in the South of Brazil and had none or incomplete or fundamental educational background (Table 1). Men and women were equally represented. 49.1% (95% CI 48.3–49.9) of the sample had one or more morbidities and 24.2% (95% CI 23.5–24.9) were multimorbid. Prevalence rates of multimorbidity increased with age, but after the age of 65 years started to level off (Table 1). 55.9% of people aged 45–59 years had one or more morbidity and 53.2% people aged 65–69 years were multimorbid (data not shown). In absolute terms, multimorbidity was 2.5-fold higher in people younger than 65 years (9920 vs 3954). Multimorbidity prevalence was higher in women than men (Table 1). 21.1% of men and 29.3% of women aged 20–24 years already had one or more morbidity (S1 Fig). Most women aged 40–44 years and most men aged 50–54 years had one or more morbidity. Half of the women and men aged 55–59 years and 65–69 years, respectively, were multimorbid. An adjusted Poisson regression model showed that multimorbidity varied across the country regions and was significantly higher in women, older people and those with lowest educational level (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Multimorbidity (≥2 conditions) in the Brazilian adult general population (n = 60202).

https://doi.org/10.1371/journal.pone.0171813.t001

Overall, 4.8% (95% CI 4.5, 5.1%) of the study population had any mental health disorder. The proportion of people with any mental health disorders increased with number of physical conditions, with up to 23.4% in people with five or more physical health problems (Table 2). However, as much as 74.6% (95% CI 71.8, 77.1%) of people with any mental health disorder had one or more chronic physical health problem. Prevalence of any mental health disorder was highest in women aged 24–44 in general and in men 45–64 and ≥85 years with four or more physical health conditions (S2 Fig). Likewise for multimorbidity prevalence rate ratios of any mental health disorder varied significantly across the country regions and were higher in women and people with lowest educational level (Table 2), but overall age were not significantly associated with having any mental health disorder.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Any mental health disorder by sex, age, educational level, country region and number of physical disorders (n = 60202).

https://doi.org/10.1371/journal.pone.0171813.t002

Fig 1 shows numbers of conditions comorbid with each specific chronic disorder. 88.3% of people with stroke had one comorbidity and as many as 66.5% had two or more comorbidities. For comparison, 62.9% of people with asthma had one comorbid disorder and 26.3% had two or more comorbidities. People with major depressive disorder had a comparable prevalence rate of comorbidity to people with severe mental health disorders such as schizophrenia, obsessive-compulsive disorder or bipolar disorder. The most prevalent comorbid condition was hypertension, any chronic back problem, high blood cholesterol and arthritis or rheumatism (S3 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Numbers of conditions comorbid with each specific chronic disorder.

Abbreviations: COPD, chronic obstructive pulmonary disorder; CVD, cardiovascular disorders; OCD, obsessive-compulsive disorder; MSK-D, musculoskeletal disorder. Proportions incorporate appropriate weights to control for the complex sample design.

https://doi.org/10.1371/journal.pone.0171813.g001

The exploratory tetrachoric factor analysis revealed three relevant patterns. The KMO coefficient was 0.83 that corresponds to a satisfactory sampling adequacy of the data. The screen plot with the parallel analysis can be found in supplementary appendix (S4 Fig). Cumulative variance after three factors retention was 99% (Factor 1: 67%, Factor 2: 21%, Factor 3: 11%). Factor loadings after oblique rotation can be found in Table 3. The first factor can be interpreted as “cardio-metabolic” (Men: 26.7% and Women: 34.5%). The second factor was named “musculoskeletal-mental” (Men: 27.0% and Women: 38.1%) and the third “respiratory” (Men: 4.8% and Women: 6.3%). 46.7% of the persons were assigned to at least one of the three multimorbidity patterns.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Factor loadings after oblique rotation [>0.25] (n = 60202).

https://doi.org/10.1371/journal.pone.0171813.t003

Discussion

Our analysis of a large, nationally representative dataset shows that overall as many as 24.2% of the community-based Brazilian adults have multimorbidity of chronic conditions, but estimates vary across age, sex, educational levels and country regions. Our overall estimate is comparable to that reported for 1751841 people registered with 314 medical practise in Scotland (23.2%) [2]–to date the largest sample used for multimorbidity analysis and perhaps one of the most cited work on epidemiology of multimorbidity—suggesting that multimorbidity in the general adult Brazilian population reached the figures of richer countries. Our estimate for the Southern region of Brazil [25.7% (95% CI 24.4, 27.0)] is comparable to the estimate previously reported for that population in a southern Brazilian city (29.1%) [20]. Likewise in Finland, India, Poland, Russia, South Africa and Spain, prevalence of multimorbidity was highest in people with lower education [13]. Previous research showed prevalence of multimorbidity is highest in older people and increases with age [2, 4, 13, 20], this pattern we found in the Brazilian population.

What is new or less described in the literature is that in Brazil people develop morbidities and multimorbidity at younger age than people in more affluent countries [2, 4] and women 10 years faster than men. Interestingly, in the Scottish population multimorbidity prevalence continued substantially growing after age of 65 years [2], in several low and middle income countries [13] and in our population multimorbidity prevalence rates levelled off after age of 65–70 years. As per previous research, women had consistently higher prevalence rates of mental health problems [2], but in our research it was not associated with age. A detailed exploration revealed possible critical periods, with highest prevalence rates of any mental health disorder at aged 24–44 years in women and 45–64 and ≥85 years in men. Overall, prevalence of any mental health disorder increased with number of physical health problems, as shown in the previous research [2]. We further found that while 1 in 4 people with one or more physical health problems had mental health comorbidity, 3 in 4 people with mental health problem had comorbid physical health problem. Whilst our findings confirm that overall hypertension is the most prevalent comorbidity in the middle-income countries [13, 20], we further found that it is people with stroke that have the highest prevalence of comorbidity and people with asthma have the lowest. The identified three patterns of multimorbidity (“cardio-metabolic”, “musculoskeletal-mental” and “respiratory”), resemble those found by Garin and colleagues [13] across middle and low-income countries (named: “metabolic”, “mental-articular” and “respiratory”) using a similar methodology.

We analysed a large sample of general adult population representative to macro-regions of Brazil, the world’s fifth largest population, including a range of chronic diseases and demographic factors. The PNS survey was very well designed and had a good response rate [25, 26]. One study limitation is unobtainability of data on socioeconomic classes. The PNS questionnaire included questions on income, but this data was not made available to the public domain. Socioeconomic classes were previously found to be significantly associated with multimorbidity in several countries including Brazil [2, 13, 20], albeit the evidence is not consistent across all countries [13].

A major methodological consideration is the definition and scope of chronic condition used. A systematic review of multimorbidity studies found that the most common source of data were self-reports and the most common cut off point for multimorbidity was two conditions [44], likewise in our study. Chronic morbidity estimates based upon self-reports may lead to underestimation of multimorbidity, with one study showing that more than half of the respondents with chronic diseases failed to report at least one disease [45]. Another study suggests that patients' self-reports on selected chronic diseases are fairly accurate, with the exceptions of atherosclerosis and arthritis, where men tend to over-report stroke and underreport malignancies and arthritis and women tend to over-report malignancies and arthritis [46]. Overall, either over-reporting or under-reporting may have occurred [45, 46]. Nevertheless, we considered the PNS data most useful for the study purpose, because reliable estimates of diagnosis are unobtainable on a national level through medical records in Brazil, due to medical records remaining often non-computerized and non-integrated, common self-referrals to secondary and tertiary care services, and simultaneous access to both public and private health systems. To date no nationally representative survey in Brazil used symptom-based algorithms or measurements to assess chronic conditions.

Another issue is a possible impact of gender on likelihood of being diagnosed with a disease. The consultation behaviour of the Brazilian population is under-researched, but one large study suggests that women in Brazil may be more likely to consult then men [47]. Existing evidence on gender differences in medical consultation from other countries is mixed [48–50], and evidence comparing consultation patterns in men and women with similar morbidity is inconsistent [51, 52]. Overall, we cannot eradicate the possibility that the reported gender difference in multimorbidity was linked to gender differences in consultation patterns. Nevertheless, we attempted to minimise this issue by adjusting Poisson models for factors that could affect consultation behaviour.

Primary care plays a vital role in the prevention and control of chronic disease worldwide [53–55]. The need for strengthening multi-disease approaches to healthcare, including primary care services is well recognized [2, 9]. In more affluent countries management of multimorbidity remains challenging due to: disorganisation and fragmentation of healthcare [56]; the inadequacy of guidelines and evidence-based medicine [56]; challenges in delivering patient-centred care [56]; barriers to shared decision-making [56] and limited evidence on interventions designed to improve outcomes in people with multimorbidity [57]. In Brazil, those problems are likely to hold true due to complexity and numerous limitations of the healthcare system. Currently, the management of multimorbidity in primary care in Brazil is not an established priority. The National Program for Access and Quality Improvement in Primary Care (PMAQ) was created in 2011 [58]. Diabetes, hypertension and drug/alcohol are the chronic conditions included in the PMAQ´s quality of outcome framework [58]. Our study emphasises a need to consider expanding the focus of primary health care teams beyond those conditions to align health-care services more closely with multimorbid patients’ needs. To meet this demand effectively a substantial increase in initiatives to address fundamental research questions is needed, such as the impact of chronic non-communicable diseases on healthcare spending [59] and integration of healthcare [60], as it is happening increasingly in more affluent countries.

Conclusions

Our results challenge the existing approach focused on patients with only one disease, currently dominating clinical guidelines and the health care system in Brazil. In order to meet the healthcare needs of the diverse Brazilian population with prevalent multimorbidity of chronic conditions, a national action plan will be needed to support primary care teams in provision of person-centred, continues and collaborative care for people with multimorbidity. Our findings suggest that beneficiaries of this approach are most likely to be women for whom multimorbidity happens earlier, is more common and more frequently is associated with mental health disorder. Whilst multimorbidity is more common in older people, we showed that strategies to prevent and control chronic conditions in Brazil should not be limited to older adults.

Supporting information