Introduction
Depression is a substantial global health concern with a lifetime prevalence ranging from 1.5% to 19% (Kessler & Bromet, Reference Kessler and Bromet2013) and is the second leading cause of disability-adjusted life years (DALYs) caused by mental and substance-use disorders (Whiteford et al. Reference Whiteford, Degenhardt, Rehm, Baxter, Ferrari, Erskine, Charlson, Norman, Flaxman, Johns, Burstein, Murray and Vos2013). Moreover, physical co-morbidities can substantially add to the increased disability and burden. Recent meta-analyses have demonstrated that people with depression are at greatly increased risk of chronic physical disease such as cardiovascular disease (Pan et al. Reference Pan, Sun, Okereke, Rexrode and Hu2011), the metabolic syndrome (Vancampfort et al. Reference Vancampfort, Stubbs, Mitchell, De Hert, Wampers, Ward, Rosenbaum and Correll2015) and diabetes (Vancampfort et al. Reference Vancampfort, Correll, Galling, Probst, De Hert, Ward, Rosenbaum, Gaughran, Lally and Stubbs2016). A previous study involving over 245 000 participants from 60 countries showed that people with depression are likely to have various physical diseases and the co-occurrence of depression and a physical disease was associated with worse health outcomes (Moussavi et al. Reference Moussavi, Chatterji, Verdes, Tandon, Patel and Ustun2007).
Despite the knowledge of the heightened risk and burden of various physical co-morbidities in people with depression and vice versa, relatively little is known about multimorbidity (i.e. two or more physical health co-morbidities) in individuals with depression. Multimorbidity has emerged as a challenging public health concern associated with a range of deleterious outcomes such as functional decline (Ryan et al. Reference Ryan, Wallace, O'Hara and Smith2015), worse quality of life (Fortin et al. Reference Fortin, Lapointe, Hudon, Vanasse, Ntetu and Maltais2004), increased risk of premature mortality (Gallo et al. Reference Gallo, Hwang, Joo, Bogner, Morales, Bruce and Reynolds2016) and increased health care costs (Lehnert et al. Reference Lehnert, Heider, Leicht, Heinrich, Corrieri, Luppa, Riedel-Heller and König2011). In the largest study to date, people with depression who attended primary care in Scotland (n = 143 943) were significantly more likely to have multiple levels of co-morbidity (Smith et al. Reference Smith, McLean, Martin, Martin, Guthrie, Gunn and Mercer2014). Yet there is a relative paucity of representative information on physical health multimorbidity in people with depression outside of primary care. In addition, there is increasing recognition that depression lies on a continuum, and that those with subsyndromal depression and brief episode disorder also may experience an increased risk of various physical co-morbidities (Ayuso-Mateos et al. Reference Ayuso-Mateos, Nuevo, Verdes, Naidoo and Chatterji2010). To our knowledge, no study has investigated multimorbidity in people with subsyndromal depression or in those with an episode lasting fewer than 2 weeks (i.e. brief depressive episode).
Furthermore, there is relatively little knowledge regarding multimorbidity in low- and middle-income countries (LMICs). This is an important research gap, since increasing trends of multimorbidity in LMICs will have considerable financial implications over the next few decades if health systems are to respond appropriately to this emerging challenge (Garin et al. Reference Garin, Koyanagi, Chatterji, Tyrovolas, Olaya, Leonardi, Lara, Koskinen, Tobiasz-Adamczyk, Ayuso-Mateos and Haro2016). Furthermore, multimorbidity in depression in the context of LMICs may differ from that of high-income countries. For example, the risk for cardiometabolic diseases may differ due to limited availability of psychotropic medications which is known to increase the risk of several physical diseases (Correll et al. Reference Correll, Detraux, De Lepeleire and De Hert2015). In addition, although smoking is known to be a major risk factor for a wide range of physical diseases for people with depression in developed countries, differences in smoking rates between those with and without depression might not be so large in some LMICs (Storr et al. Reference Storr, Cheng, Alonso, Angermeyer, Bruffaerts, De Girolamo, De Graaf, Gureje, Karam and Kostyuchenko2010). Moreover, conflict, elevated rates of HIV and abject poverty further add to the risk and complexities of physical health multimorbidity among people with mental illness in LMICs (Mugisha et al. Reference Mugisha, Ssebunnya and Kigozi2016).
Given the aforementioned gaps within the literature, we aimed to assess the association between the whole depressive spectrum and multimorbidity among community individuals with depression across 43 LMICs.
Method
The World Health Survey was a cross-sectional study undertaken in 2002–2004 in 70 countries worldwide. Single-stage random sampling and stratified multi-stage random cluster sampling were conducted in 10 and 60 countries, respectively. The details of the survey have been provided elsewhere (http://www.who.int/healthinfo/survey/en/). Briefly, persons aged ⩾18 years with a valid home address were eligible to participate. Each member of the household had equal probability of being selected with the use of Kish tables. The data were collected in all countries using the same questionnaire. The individual response rate (ratio of completed interviews among selected respondents after excluding ineligible respondents from the denominator) ranged from 63% (Israel) to 99% (the Philippines) (Moussavi et al. Reference Moussavi, Chatterji, Verdes, Tandon, Patel and Ustun2007). The data were nationally representative in all countries with the exception of China, Comoros, the Republic of Congo, Ivory Coast, India and Russia. Ethical approval to conduct this survey was obtained from ethical boards at each study site. Sampling weights were generated to adjust for non-response and the population distribution reported by the United Nations Statistical Division. Informed consent was obtained from all participants.
Physical health conditions
A total of nine physical conditions were assessed. Arthritis, asthma and diabetes were based on self-reported lifetime diagnosis. For angina, in addition to a self-reported diagnosis, a symptom-based diagnosis based on the Rose questionnaire was also used. Chronic back pain was defined as having had back pain (including disc problems) every day during the last 30 days. Visual impairment was defined as having extreme difficulty in seeing and recognizing a person that the participant knows across the road (i.e. from a distance about 20 m) (Freeman et al. Reference Freeman, Roy-Gagnon, Samson, Haddad, Aubin, Vela and Zunzunegui2013). A validity study showed that this response probably corresponds to World Health Organization definitions of visual impairment [20/60 or 0.48 logMAR (logarithm of the minimum angle of resolution)] (Freeman et al. Reference Freeman, Roy-Gagnon, Samson, Haddad, Aubin, Vela and Zunzunegui2013). The participant was considered to have hearing problems if the interviewer observed this condition at the end of the survey. Edentulism was assessed by the question ‘Have you lost all your natural teeth?’. Those who responded affirmatively were considered to have edentulism. Finally, a tuberculosis diagnosis was based on past 12-month symptoms and was defined as: (1) having had a cough that lasted for 3 weeks or longer; and (2) having had blood in phlegm or coughed up blood. We calculated the total number of these conditions while allowing for one missing variable in order to retain a larger sample size. Multimorbidity was defined as having at least two conditions, in line with previously used definitions (Fortin et al. Reference Fortin, Stewart, Poitras, Almirall and Maddocks2012).
Severity of depressive symptoms
The severity of depressive symptoms was established based on the individual questions of the World Mental Health Survey version of the Composite International Diagnostic Interview (CIDI), which assessed the duration and persistence of depressive symptoms in the past 12 months (Kessler & Ustun, Reference Kessler and Ustun2004). Following the algorithms used in a previous World Health Survey publication (Ayuso-Mateos et al. Reference Ayuso-Mateos, Nuevo, Verdes, Naidoo and Chatterji2010), four mutually exclusive groups were established based on the International Classification of Diseases, tenth revision Diagnostic Criteria for Research (ICD-10-DCR) where criterion B referred to symptoms of depressed mood, loss of interest and fatigability. The algorithms used to define the four groups were the following:
Depressive episode group: at least two criterion B symptoms together with a total of at least four depressive symptoms lasting 2 weeks most of the day or all of the day.
Brief depressive episode group: same criteria as depressive episode above but duration did not meet the 2-week duration criterion.
Subsyndromal depression: at least one criterion B symptom together with the total number of symptoms being three or fewer. The criteria of duration of at least 2 weeks and presence of symptoms during most of the day had to be met.
No depressive disorder group: none of the above.
Any depression is hereby defined as any of depressive episode, brief depressive episode or subsyndromal depression.
Other variables
Information was also examined on age, sex, wealth and education. Principal component analysis based on 15–20 assets was conducted to establish country-wise wealth quintiles. Education was categorized as: no formal education, primary education, secondary or high school completed, or tertiary education completed.
Statistical analysis
Data from 69 countries were publicly available. Countries without any sampling information (10 countries – Austria, Belgium, Denmark, Germany, Greece, Guatemala, Italy, the Netherlands, Slovenia and the UK) were excluded. Of the remaining 59 countries, 10 (Finland, France, Republic of Ireland, Israel, Luxembourg, Norway, Portugal, Sweden, Spain and the United Arab Emirates) were subsequently excluded as they were high-income countries. Furthermore, Turkey was excluded owing to missing information on depression and a further five countries (Congo, Mali, Mexico, Slovakia, Swaziland) were omitted as >25% of the data on multimorbidity was missing. The final sample thus included 43 countries which consisted of 19 low-income and 24 middle-income countries according to the World Bank classification in 2003. Although five countries in the final analytical sample were not nationally representative, we included these countries in the analysis as preliminary analyses showed that the exclusion or inclusion of these countries does not appreciably change the results.
The statistical analysis was performed with Stata 14.1 (Stata Corp LP, USA). Descriptive analysis was performed to calculate the prevalence of each physical condition and number of physical health conditions by each type of depression. Tetrachoric correlations between each pair of physical health condition were calculated for those with any type of depression. Furthermore, the prevalence of co-occurring physical conditions for individual somatic conditions was also calculated. Multivariable logistic regression analysis was performed to assess the association between depression types (independent variable) and multimorbidity (dependent variable) while adjusting for age (continuous variable), sex, education, wealth and country. Age group-stratified analyses (18–44, 45–64, ⩾65 years) were also conducted. To adjust for country, dummy variables for each country were included in the models, following the methods used in previous World Health Survey publications (Koyanagi et al. Reference Koyanagi, Oh, Stickley, Haro and DeVylder2016). We also conducted country-wise logistic regression analyses to assess the association between any depression and multimorbidity while adjusting for sex and age. A pooled estimate was obtained by combining the estimates for each country into a random-effects meta-analysis. This was done to assess the robustness of the analysis by applying two different types of analytical methods and also to evaluate the degree of generalizability of our findings across countries.
Less than 10% of the data were missing for all variables used in the analysis with the exception of tuberculosis (12.6%). No attempts to impute missing data were made because we had no information about whether these data were missing at random. The sample weighting and the complex study design were taken into account in the analyses. Results from the logistic regression models are presented as odds ratios (ORs) with 95% confidence intervals (CIs). The level of statistical significance was set at p < 0.05.
Results
The final sample size was 190 593 people from 43 LMICs. Within the final sample, females constituted 50.7% of the sample while the mean age was 38.4 (s.d. = 16.0) years. The prevalence of subsyndromal depression, brief depressive episode and depressive episode were 2.5, 2.9 and 6.8%, respectively. The prevalence of physical conditions ranged from 1.4% (visual impairment) to 15.5% (angina) (Table 1). Compared with those with no depression, the prevalence of each physical condition was much higher among those with all types of depression (Table 1). Across almost all of the physical health co-morbidities, there was an incremental increase in the prevalence of each physical health co-morbidity from subsyndromal depression and brief depressive episode to those experiencing a depressive episode (Table 1). Overall, multimorbidity was present in 13.5% of the entire sample, with two, three and four or more physical health conditions present in 7.4, 2.4 and 0.9% of people without depression v. 17.7, 9.1 and 4.9% in people with any depressive episode, respectively (Table 1). The prevalence of multimorbidity in those aged 18–44, 45–64 and ⩾65 years was 6.2, 22.0 and 48.1% (χ2 test, p < 0.0001), respectively, while the corresponding figures for any depression were 10.4, 15.7 and 20.0% (χ2 test, p < 0.0001).
Table 1. Prevalence of physical health conditions (overall and by different types of depression) a

Data are given as unweighted n/N and weighted percentage.
a Data on prevalence for depression are column percentage (i.e. the prevalence of each physical health condition or number of health conditions among those in that depression category).
b Any depression refers to having subsyndromal depression or brief depressive episode or depressive episode.
Relationship between physical health co-morbidities among people with depression
Among those with any type of depression, a particularly strong correlation was observed for visual impairment and hearing problems, edentulism, as well as arthritis and angina (Table 2). This was also reflected, for example, in the high prevalence of hearing problems among those with visual impairments and edentulism (Table 3). A particularly high prevalence of diabetes was observed for those with visual impairments and edentulism.
Table 2. Tetrachoric correlations of physical health conditions among those with any depression a

a Any depression refers to having subsyndromal depression or brief depressive episode or depressive episode.
Table 3. Prevalence of other physical conditions for each physical condition among those with any depression a

Data are given as unweighted n/N and weighted percentage.
a Any depression refers to having subsyndromal depression or brief depressive episode or depressive episode.
The association between different types of depression and multimorbidity (i.e. having two or more physical health conditions) estimated by multivariable logistic regression is shown in Table 4. In the overall sample, compared with those with no depressive disorder, subsyndromal depression, brief depressive episode and depressive episode were significantly associated with 2.62, 2.14 and 3.44 times higher odds for multimorbidity, respectively (p < 0.0001 for all elaborations). Similar estimates were obtained for the different age groups but the association was stronger in the younger age group (Table 4).
Table 4. Association between different types of depression and multimorbidity (outcome) estimated by multivariable logistic regression a

OR, Odds ratio; CI, confidence interval.
a Models are adjusted for age, sex, wealth, education and country.
Country-wide meta-analysis of depression and multimorbidity
Country-wide meta-analysis adjusted for age and sex demonstrated that overall, across 43 countries, depression is associated with a 3.26 times increased odds of physical health multimorbidity. Moreover, the association between any depression and physical health multimorbidity was significant in all the countries studied with the exception of Vietnam. The findings from Vietnam should be interpreted with caution as there were only 33 cases of any depression. Particularly high odds of depression and physical health multimorbidity were evident in China (OR 8.84), Laos (OR 5.08), Ethiopia (OR 4.99), the Philippines (OR 4.81) and Malaysia (OR 4.58), among others (Fig. 1).

Fig. 1. Country-wise association between any depression (independent variable) and multimorbidity (dependent variable) estimated by logistic regression adjusted for sex and age. ‘Any depression’ refers to having subsyndromal depression or brief depressive episode or depressive episode. OR, Odds ratio; CI, confidence interval.
Discussion
To the best of our knowledge, the current study is the first multinational research to investigate the association of clinical depression, subsyndromal depression and brief depressive episode with physical health multimorbidity. Compared with those with no depressive disorder, subsyndromal depression, brief depressive episode and depressive episode were significantly associated with 2.62, 2.14 and 3.44 times higher odds for multimorbidity, respectively. Given that physical health multimorbidity can greatly increase the risk of mortality (Di Angelantonio et al. Reference Di Angelantonio, Kaptoge, Wormser, Willeit, Butterworth, Bansal, O'Keeffe, Gao, Wood, Burgess, Freitag, Pennells, Peters, Hart, Haheim, Gillum, Nordestgaard, Psaty, Yeap, Knuiman, Nietert, Kauhanen, Salonen, Kuller, Simons, van der Schouw, Barrett-Connor, Selmer, Crespo, Rodriguez, Verschuren, Salomaa, Svardsudd, van der Harst, Bjorkelund, Wilhelmsen, Wallace, Brenner, Amouyel, Barr, Iso, Onat, Trevisan, D'Agostino, Cooper, Kavousi, Welin, Roussel, Hu, Sato, Davidson, Howard, Leening, Rosengren, Dorr, Deeg, Kiechl, Stehouwer, Nissinen, Giampaoli, Donfrancesco, Kromhout, Price, Peters, Meade, Casiglia, Lawlor, Gallacher, Nagel, Franco, Assmann, Dagenais, Jukema, Sundstrom, Woodward, Brunner, Khaw, Wareham, Whitsel, Njolstad, Hedblad, Wassertheil-Smoller, Engstrom, Rosamond, Selvin, Sattar, Thompson and Danesh2015), health care costs (Lehnert et al. Reference Lehnert, Heider, Leicht, Heinrich, Corrieri, Luppa, Riedel-Heller and König2011) and work absenteeism, our results demonstrate the considerable health strain evident within the depression spectrum. However, of note, it is important to illustrate that the directionality or conjecture of the relationships observed cannot be determined due to the cross-sectional nature of the study.
Interestingly, whilst our data found that depression in all age ranges is associated with increased odds of experiencing physical health multimorbidity, people aged between 18 and 44 years appear to be at particular risk. The precise reasons for this stronger association between depression and physical health multimorbidty in the younger age group are not clear but clearly are of concern. One possible reason is that the presence of multiple chronic physical health conditions may result in less physical activity and social isolation (Schuch et al. Reference Schuch, Vancampfort, Firth, Rosenbaum, Ward, Reichert, Bagatini, Bgeginski and Stubbs2016), which may in turn lead to depression. This may have a particularly detrimental effect on mental health in the younger age group as they are usually more active. In addition, this younger age bracket is also highly affected by conflict, HIV and economic stress in LMICs which may account for the heightened odds of depression and physical health multimorbidity (Mugisha et al. Reference Mugisha, Ssebunnya and Kigozi2016). Nonetheless, this finding adds further impetus to the calls for early intervention efforts to prevent the cumulative burden of physical health co-morbidity at later ages. Within our analysis, we also observed some interesting variations in the strength of the relationship between depression and physical health multimorbidity. For instance, China had a considerably elevated OR (8.84) which was much higher than other countries such as Senegal (OR 1.89). The precise reasons for this are not clear, nor can we fully consider this from the available dataset. However, cultural aspects may play an important role and this warrants careful consideration in future research.
Strategies to deal with physical health multimorbidity are urgently needed among people with depressive illness, particularly targeting the earlier stage of the illness (Ghio et al. Reference Ghio, Gotelli, Cervetti, Respino, Natta, Marcenaro, Serafini, Vaggi, Amore and Murri2015). The importance of integrating physical health multimorbidity into clinical guidelines for people with depression is needed at every level of health care systems including primary, secondary and tertiary care. The provision and support of self-care management strategies including advice on a healthy and active lifestyle, prioritizing the prevention of chronic conditions and avoiding fragmented care have been underlined previously (Barnett et al. Reference Barnett, Mercer, Norbury, Watt, Wyke and Guthrie2012). Moreover, there is also a need to recognize that mental health care co-morbidity screening among people with physical health multimorbidity is also needed. Clearly health care system integration is required and existing health care models need to adapt to the high multimorbidity rates, particularly with the management of depression and physical health multimorbidity across LMICs, where all levels of care must be carefully planned in the context of economic restraints. The Innovative Care for Chronic Conditions framework (Oni et al. Reference Oni, McGrath, BeLue, Roderick, Colagiuri, May and Levitt2014) developed by the World Health Organization provides a roadmap to cope with chronic conditions in developing countries but the concept of multimorbidity provides a framework for the organization of services within primary and mental health care settings. Health care systems in LMICs including policy makers should embrace the physical health needs in the management of depression across the entire health care system. An important environmental barrier in the care of people with depression in LMICs is the lack of integrated mental and medical services and the poorly developed community-based psychiatric services (Mugisha et al. Reference Mugisha, Ssebunnya and Kigozi2016). Closer integration of primary care and mental health in these countries is needed, but without obscuring the responsibility for physical health screening at key periods, such as upon admission to a mental health hospital or prior to starting psychotropic medication. We suggest that screening for physical health multimorbidity at these times should be the responsibility of the treating mental health professional, although clearly all health care staff need to be aware of the heightened mental health and physical health multimorbidity relationship. However, it should be acknowledged that many mental health providers appear not to ask about medical issues or test for them because of a lack of consideration of medical co-morbidities, time pressures, uncertainty of responsibility or low resources to address the problem. For example, in a South African study (Ludwick & Oosthuizen, Reference Ludwick and Oosthuizen2009), 90% of the mental health providers believe that patients with psychiatric disorders are being monitored for medical co-morbidities to a similar degree (if not higher) than the rest of the general population. The mechanisms leading to this negative ascertainment bias, in which somatic diseases are not properly recognized among those with mental disorders, remain incompletely elucidated. Both conscious (e.g. stigma of health care providers towards patients with mental disorders), but also unconscious aspects may play a role (Ludwick & Oosthuizen, Reference Ludwick and Oosthuizen2009). In addition, a lack of knowledge is exemplified by the observation that only 10% of mental health providers are aware that people with mental disorders die earlier than the general population (Ludwick & Oosthuizen, Reference Ludwick and Oosthuizen2009). Therefore, first of all, there is a clear need to increase awareness of the importance of physical health needs of patients with depression among mental health providers in LMICs.
Continued medical education, which is a common practice in LMICs (Mugisha et al. Reference Mugisha, Ssebunnya and Kigozi2016), could be used to inform health providers on the importance of assessing physical health risks in people with depressive episode. However, continued medical education in LMICs is often funded by donors who tend to prioritize issues related to communicable diseases or maternal and child health. Therefore, demonstrating the importance of physical health multimorbidity with depression to such groups may be key to ensure it is adequately covered. Policy makers should be made aware that investment in continued medical education and in screening for physical health risks could optimize mental and physical health improvements. However, effective monitoring of metabolic risks is not sufficient on its own, as appropriate treatment is also mandatory. Notwithstanding that evidence indicates that collaborative care models are cost-effective for the management of depression associated with co-occurring somatic illness in primary care, the implementation of this strategy remains an unmet need in LMICs. Disease surveillance of the mental health and physical health multimorbidity is key to assess disease burden and trajectory in a population. However, health management systems and informatics in LMICs may not be capable of recording multiple diseases across health care systems.
Limitations and strengths
The current findings should be interpreted in light of some limitations. First, the study is cross-sectional; therefore precise developmental trajectories of depression and physical health multimorbidity cannot be established. Nonetheless, regardless of directionality, the public health message remains the same. In addition, the diagnosis of subsyndromal depression, brief depressive episode and depressive episode was not assessed by a clinical interview. Moreover, the self-report measure may have led to a bias in the diagnosis of the medical conditions particularly in older subjects. Furthermore, due to poor/incomplete data, we were not able to investigate the impact of factors such as obesity or alcohol use on the relationship between depression and multimorbidity. Future research should attempt to assess the degree to which these factors contribute to this relationship. Nonetheless, the strengths of the study include the large sample size and the multinational scope, including most regions of the world, but in particular LMICs in Africa, Latin America, Asia and Eastern Europe.
In conclusion, we found a consistent elevated relationship between physical health multimorbidity and depression. Given this, people with subsyndromal depression, brief depressive episode and depressive episode in LMICs should be screened for the presence of somatic health conditions and vice versa. Future studies are needed to assess the impact of global preventive and therapeutic strategies targeting multimorbidity in this vulnerable population.
Acknowledgements
This article received no specific funding or grant. B.S. receives funding from the National Institute for Health Research Collaboration for Leadership in Applied Health Research & Care Funding scheme. The views expressed in this publication are those of the author(s) and not necessarily those of the National Health Service, the National Institute for Health Research or the Department of Health. A.K.’s work was supported by the Miguel Servet contract financed by the CP13/00150 and PI15/00862 projects, integrated into the National R + D + I and funded by the ISCIII – General Branch Evaluation and Promotion of Health Research – and the European Regional Development Fund (ERDF-FEDER). D.V. is funded by the Research Foundation – Flanders (FWO – Vlaanderen).
B.S. and A.K. conceived the study idea. A.K. led on data analysis with B.S. providing input. B.S., A.K. and D.V. wrote the first draft of the manuscript. All authors provided critical comments to the manuscript. All authors approved the final manuscript.
Declaration of Interest
None.