Abstract
Importance: The simultaneous presence of multiple conditions in one patient (multi-morbidity) is a key challenge facing primary care.
Objective: The purpose of this study was to determine the prevalence of multi-morbidity and to document changes in prevalence during the last 25 years.
Design/Setting: Cross-sectional study using multiple years (1988–2014) of the National Health and Nutrition Examination Survey (NHANES) were analyzed.
Setting: Multiple years (1988 to 2014) of the National Health and Nutrition Examination Survey (NHANES) from the United States were analyzed.
Participants: Noninstitutionalized adults.
Main Outcomes and Measures: Number of chronic conditions per individual analyzed by age, race, gender, and socioeconomic factors.
Results: A total of 57,303 individuals were surveyed regarding the presence of multi-morbidity in separate surveys spanning 1988–2014. The overall current prevalence in 2013–2014 of >2 morbidities was 59.6% (95% CI 58.1%–61.1%), 38.5% had 3 or more, and 22.7% had 4 or more morbidities, which was significantly higher than in 1988 (45.7%, 95% CI 43.5%–47.8%, with >2 morbidities). Among individuals with 2 or more morbidities, 54.1% have obesity compared to 41.9% in 1988. Among adults age >65, prevalence was 91.8% for 2 or more morbidities. Whites and Blacks had significantly higher prevalence (59.2% and 60.1%) than Hispanic or “other” race (45.0%, P < .0001). Women (58.4%) had more current multi-morbidities (>2) than men (55.9%, P = .01).
Conclusions and Relevance: Multimorbidity is common and has been increasing over the last 25 years. This finding has implications for public health policy and anticipated health costs for the coming years.
- Cross-Sectional Studies
- Health Care Costs
- Health Policy
- Multimorbidity
- Nutrition Surveys
- Prevalence
- Primary Health Care
- Socioeconomic Factors
The simultaneous presence of multiple conditions in one patient (multimorbidity) is a key challenge in primary care. Multimorbidity adds to the complexity of care and threatens the quality, coordination, continuity, and safety of care in the United States health care system and elsewhere. Despite the seriousness and far-reaching impacts of this phenomenon, characterization of this population in recent studies has focused on older populations, include a limited number of chronic conditions, and often do not include obesity as a chronic condition.1⇓⇓⇓⇓⇓⇓–8 The burden on patients with multimorbidity is considerable and is associated with increased mortality.9⇓⇓–12 A recent meta-analysis by Nunes and colleagues10 included 5806 multimorbidity studies, and mortality (26 studies were included) demonstrated a hazard ratio of 1.73 (95% CI, 1.41–2.13) and 2.72 (95% CI, 1.81–4.08) for people with 2 or more and 3 or more morbidities, respectively.
In addition, heterogeneity in the included conditions of the studies was high, and obesity was not always included in the list of comorbidities, potentially underestimating the prevalence of multimorbidity.12–13 Including obesity in multimorbidity estimates is also crucial due to the well-studied link between obesity and a variety of complications, including diabetes, heart disease, cancer, and many others.14⇓⇓⇓⇓–19 Kivimaki and colleagues20 have documented considerably increased cardiovascular events in obese versus nonobese cohorts in a pooled analysis of 16 cohort studies.
The purpose of this study was to determine the current prevalence of multimorbidity by using 11 common conditions, including obesity, and to examine trends in prevalence during the past 25 years. A secondary objective was to examine age, gender, race, and socioeconomic factors associated with multimorbidity prevalence.
Methods
Study Population
The National Health and Nutrition Examination Survey (NHANES) are serial, cross-sectional, stratified, multistage probability surveys designed to assess the health and nutrition status among the US civilian, noninstitutionalized population. The surveys were conducted by the National Center for Health Statistics, the data were collected on participants' demographic characteristics, nutrition, health, and diet through interviews in participants' homes, and medical examinations were conducted in a mobile examination center. All participants completed written informed consents, and protocols for conducting the NHANES study were approved by the Center for Disease Control and Prevention Institutional Review Board. Details on survey design and response rates can be found on the NHANES website.21
The present study combined NHANES III, which was conducted between 1988 and 1994,22 and the continuous NHANES from 1999 to 2014, with data released in 2-year cycles.22
Study Participants
Participants aged 20 years or older, with nonzero weights (not nonrespondents) were included in the study sample. Of the 57,303 participants included in the study sample, there were 16,573 from NHANES III and 40,730 from NHANES 1999 to 2014.
Multimorbidity
Multimorbidity, defined as the presence of 2 or more chronic conditions in a person, was the primary outcome of the study. Eleven chronic conditions were selected based on their clinical relevance and the availability of the NHANES data: cardiovascular disease, chronic obstructive pulmonary disease, chronic kidney disease, asthma, arthritis, cancer, stroke, hypertension, hyperlipidemia, diabetes, and obesity.
NHANES collects self-reports of diagnosis by a doctor for health conditions by asking a participant “have you ever been told by a doctor that you have that condition?” Participants were classified as having asthma, arthritis, stroke, and cancer, if the participant gave a positive answer to the question regarding these conditions. Participants were classified as having cardiovascular disease if they answered yes to having at least 1 of the following heart conditions: congestive heart failure, coronary heart disease, or heart attack. Although all 3 heart conditions were asked about in NHANES 1999 to 2014, participants in NHANES III were asked only about 2 of these conditions, congestive heart failure and heart attack. Participants were classified as having chronic obstructive pulmonary disease if they answered yes to having emphysema or chronic bronchitis.
Participants were classified as having hypertension, hyperlipidemia, or diabetes if they gave a positive answer to the self-reported question or had an individual medical measurement equal or greater than the recommended threshold. For example, a participant would be identified as having diabetes if he/she answered “yes” to the question regarding diabetes or had a measured hemoglobin A1c ≥6.5%. The hemoglobin A1c cutoff was determined using the consistent standard set by the American Diabetes Association that is summarized in their clinical guidelines.23
Blood pressure cutoffs for hypertension were greater than 140 mmHg for systolic blood pressure or 90 mmHg for diastolic blood pressure.24 The cholesterol cutoff for determining hyperlipidemia was greater than 200 mg/dL of the total cholesterol, based on the Adult Panel III guidelines.25
Participants were classified as obese if they had a body mass index ≥ 30 kg/m2.
There was no self-report of diagnosis question for chronic kidney disease in NHANES. To identify participants with chronic kidney disease, we estimated level of kidney function from the estimated glomerular filtration rate (eGFR), which was calculated from recalibrated serum creatinine26 by using the Chronic Kidney Disease Epidemiology Collaboration creatinine equation.27 Specifically, eGFR = 141 × min(Scr/k, 1)a × max(Scr/k, 1)−1.209 × 0.993Age × 1.018 [if female] × 1.159 [if black]; with Scr = serum creatinine (mg/dL); k = 0.7, and a = −0.329 if female; k = 0.9 and a = −0.411 if male. As lower values of eGFR correspond with diminished kidney function, participants with a value of eGFR <60 mL/min/1.73 m2 were identified as having chronic kidney disease.
Multimorbidity was categorized as ≥2 morbidities, ≥3 morbidities, and ≥4 morbidities.
Covariates
Other demographic characteristics extracted for this study included age, gender, race, and socioeconomic status (education level, health insurance status, and ratio of family income to poverty). Age was divided into 3 groups: 20 to 44 years, 45 to 64 years, and 65 years or older. Race was combined into 4 groups of non-Hispanic white, non-Hispanic black, Hispanic, and other race. Participants' education level was grouped into 2 categories of “<high school” and “≥high school.” The ratio of family income to poverty was recoded as “above poverty” for greater or equal to 1.0 and “under poverty” for less than 1.0. Participants' health insurance status was defined as “yes” for having health insurance and “no” for not having health insurance.
Statistical Analysis
All data analyses were performed with SAS package version 9.3 (SAS institute Inc., Cary, NC). To account for the complex survey design (including oversampling), survey nonresponse, and poststratification, we incorporated appropriate sampling weights and SAS survey analysis procedures following NHANES analytic and reporting guidelines.28 Two year weights for NHANES 1999 to 2014 and 6-year weights for NHANES III were used for the prevalence estimate of individual cohorts. For trend analysis, we used the combined 6-year weights (NHANES III) and 16-year weight for NHANES 1999 to 2014. Missing data were assumed to be missing at random. To account for the confounding effect of age, age-standardized prevalence levels of multimorbidity (≥2 morbidities, ≥3 morbidities, and ≥4 morbidities) were estimated and compared using F tests for overall samples and subsamples were stratified by NHANES cycles, race, gender, education level, health insurance status, and poverty level. The US 2010 Census population of adults aged 20 years or older was used for the calculation of the age-group weights (20 to 44 years, weight 0.5114; 45 to 64 years, weight 0.3114; and 65 years or older, weight 0.1772).29 Logistic regression was performed to assess linear trends in levels of multimorbidity across NHANES cycles overall and by demographic and socioeconomic status. P values for the trend analysis were calculated by regressing the levels of multimorbidity on the median year of the survey cycle. Statistical significance was determined if a 2-sided P value <.05.
Results
The prevalence of multimorbidity by demographic characteristics in NHANES 2013 to 2014 is presented in Table 1. Among the total sample of 5541 participants in the 2013 to 2014 cycle, 59.6% (95% CI, 58.1%–61.1%) had ≥2 morbidities, 38.5% (95% CI, 36.3%–40.6%) had ≥3 morbidities, and 22.7% (95% CI, 21.1%–24.3%) had ≥4 morbidities.
Age-Standardized Prevalence of Multimorbidity in Participants 20 Years or Older, Stratified by Age, Sex, Race, Poverty, Education, and Insurance Status for National Health and Nutrition Examination Survey (NHANES) 2013–2014
Compared with the groups aged 45 to 64 years and 20 to 44 years, the weighted prevalence of ≥2 morbidities was higher in those aged 65 years or older (91.8% vs 70.6% vs 37.5%, P < .0001). A similar significant difference between the age groups was also found in the prevalence of individuals with ≥3 morbidities (76.5% vs.47.7% vs.15.3%, P < .0001) and ≥4 morbidities (55.6% vs 26.4% vs 6.0%, P < .0001).
There was higher age-standardized prevalence at all levels of multimorbidity in female participants than in male participants (58.4% vs 55.9%, P = .01 for ≥2 morbidities; 38.4% vs 33.8%, P = .0002 for ≥3 morbidities; and 23.6% vs 18.4%, P < .0001 for ≥4 morbidities).
Across all 3 levels of multimorbidity, the age-standardized prevalence was consistently higher in non-Hispanic white and non-Hispanic black participants than Hispanic participants or participants of other race. The age-standardized prevalence of all levels of multimorbidity was similar among different education groups. Participants with health insurance had higher prevalence of ≥2 and ≥3 multimorbidities than their counterparts without health insurance. There was a lower prevalence of ≥4 multimorbidities in participants “above poverty” compared with those in the “under poverty” group.
Tables 2, 3, and 4 summarize the trends in multimorbidity prevalence between 1988 and 2014. The weighted overall prevalence of ≥2 multimorbidities, ≥3 multimorbidities, and ≥4 multimorbidities significantly increased from 45.7%, 24.6%, and 12.0% in 1988 to 1994 to 59.6%, 38.5%, and 22.7% in 2013 to 2014 (P < .0001 for trend for all 3 levels) (as summarized in Figure 1). Significant increases in multimorbidity prevalence over the study period were seen in all levels of multimorbidity and for all age, gender, race, health insurance status, poverty level, and education level groups except other race. Although not significant, there was a decreasing trend in multimorbidity prevalence for other race.
Age-Standardized Prevalence of 2 or More Multimorbidities for Participants Aged 20 Years or Older, Stratified by Age, Sex, Race, Poverty, Education, and Insurance Status for National Health and Nutrition Examination Survey (NHANES) 1988–2014
Weighted Prevalence of 3 or More Multimorbidities for Participants Aged 20 Years or Older for National Health and Nutrition Examination Survey (NHANES) 1988–2014
Weighted Prevalence of 4 or More Multimorbidities for Participants Aged 20 Years or Older for National Health and Nutrition Examination Survey (NHANES) 1988–2014
Age-standardized trends in multi-morbidity prevalence for participants 20 years or older from NHANES 1988–2014 by number of comorbidities.
Figure 2 illustrates the prevalence of each individual morbidity condition in the cohorts included in the study. Obesity experienced the largest increased trend of any condition across the study time frame (P < .0001).
Age-standardized prevalence of various chronic conditions in participants 20 years or older from NHANES 1988–2014.
Discussion
The current findings document the high and growing prevalence of multimorbidity in adults in the United States. Overall, we observed over half of all adults (59.6%) aged 20 and older have 2 or more multimorbidities, a proportion that has steadily increased from 45.7% in the 1988 to 1994 survey period. The prevalence was highest in people aged 65 years or older (91.8%) and consistently higher in females than males.
These trend results are consistent with recent data from the Centers for Disease Control and Prevention showing a high prevalence of comorbidity in people with chronic conditions. For example, data from the National Health Interview Survey showed that 49% of people with heart disease also had doctor-diagnosed arthritis.30 Recent Centers for Disease Control and Prevention data also showed that 25% of adults had at least 2 chronic conditions (out of 10 possible conditions).31 Dugolf and colleagues32 documented prevalence among Medicare beneficiaries and concluded that more than two-thirds of older adults have at least 2 chronic conditions.
The current study results show a higher prevalence than seen in other similar studies that is likely secondary to our selection of chronic conditions, notably obesity. Obesity is associated with a large number of pathologic processes and risks, including metabolic syndrome, vascular disease, cancer, oxidative stress, inflammation, as well as many others. Due to the considerable morbidity of obesity and its impact on a variety of health systems, we felt it was important to include it as a chronic condition rather than a control factor in the multimorbidity calculations for the current study.15⇓⇓⇓–19
Similar to the current study's observation of increasing multimorbidity, this trend has been seen in other cohorts and other countries. Oostrum and colleagues33 examined multimorbidity trends from 2001 to 2011 and saw increases in multimorbidity, but published much lower rates of multimorbidity (14.3% to 17.5%, P < .01), despite including 28 conditions seen in general practice. Their list included heart, lung, and mood disorders, as well as many others. A study in Canada by Pefoyo and colleagues34 reported a multimorbidity rate that was increasing (24.3%), but was still much lower than in the current study.
In US studies on a state level, similar patterns to the current study have been documented. Rocca and colleagues35 have studied a Minnesota cohort and reported similar trends for age and sex as in the current study.
Their overall rate of multimorbidity using 20 conditions was 77.3% for ages 65 years and older, compared with our finding of 91.8% in participants over 65, but their study did not include obesity as one of the conditions.
The increase in multimorbidity over time suggests a worsening of the disease burden faced by individuals of all demographic characteristics. Over 91% of people over 65 are dealing with at least 2 serious chronic conditions or risk factors, and many are facing 4 or more. Prince and colleagues36 have recently reviewed the chronic disease burden among older people and concluded that it is a global problem and an epidemic. Furthermore, care of older adults with cardiovascular conditions is significantly complicated by the concurrent comorbidity burden that so frequently accompanies them.37
Possible explanations for the increasing prevalence of multimorbidity have been documented in the literature on numerous occasions, including unhealthy diet patterns, infrequent regular physical activity, smoking, and socioeconomic factors.38⇓⇓⇓⇓⇓–44 Other possible explanations are the prevalence of health disparities and the ease and regularity of access to primary care, which would lead to increased diagnosis.45–46 The current study observed that much of the increasing trend in multimorbidity was likely due to the significant increase in obesity.
The association between trends and morbidities in people with insurance is complex and has been the subject of numerous studies, including 24 recent cross-sectional studies.47
Extensive further study will be needed to determine the roots of multimorbidity differences in populations and the impact on outcomes and disability.48
The association seen in the current study between having insurance and more comorbidities may be a consequence of several possible factors, including that insured people have easier health care access and may more frequently be told a diagnosis. Underdiagnosis of poorer individuals and uninsured populations also may be contributing. The local physical/geographic environment, insurance copays, regional variation, and many other factors may be contributing to this insurance socioeconomic equation and need further research.
This study has several limitations, including possible misclassification, consistency of data reported over cohort years, and cross-sectional data collection. Misclassification is a concern due to the reliance on self-reporting for the determination of several of the chronic conditions. Participants were considered to have the specific chronic condition by either a doctor-diagnosed history or by reaching the threshold for certain conditions, even if not formally diagnosed, such as blood pressure >140/90 or cholesterol >200. However, classification standards were consistently applied across the NHANES cohorts in the current study.
In addition, the comorbidities included in this study were limited because all selected conditions had to be included in each year cohort of the general NHANES questionnaire. For example, depression, anxiety, opioid addiction, and other mental health conditions known to be associated with morbidity have not been included consistently for all adult age groups in the NHANES cohorts over the period of this study, thus making it likely that we have underestimated multimorbidity. Specifically, opioid overuse or abuse data were not routinely collected even though it is recognized as a significant problem and growing contributor to premature mortality.49
Another limitation is that this study population consists of a series of cross-sectional studies, and, thus, the study is examining different people at each interval and does not represent the course of chronic disease in any individual.
In conclusion, multimorbidity for the 11 selected conditions is highly prevalent and has increased over the past 25 years. Obesity is a significant contributor to the trend. Public health leaders and policy makers should be attentive to these trends when designing policies and interventions to improve the public's health. Further research is needed to determine which interventions would be most helpful in addressing people with multimorbidity.
Notes
This article was externally peer reviewed.
Funding: Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award 2U54GM104942–02. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest: none declared.
To see this article online, please go to: http://jabfm.org/content/31/4/503.full.
- Received for publication January 8, 2018.
- Revision received April 3, 2018.
- Accepted for publication April 3, 2018.
References
In this issue
Jump to section
Related Articles
Cited By...
- The Scope of Multimorbidity in Family Medicine: Identifying Age Patterns Across the Lifespan
- What is the impact of multimorbidity on out-of-pocket healthcare expenditure among community-dwelling older adults in Ireland? A cross-sectional study
- Temporal Trends in Racial and Ethnic Disparities in Multimorbidity Prevalence in the United States, 1999-2018
- Primary Care Variation in Rates of Unplanned Hospitalizations, Functional Ability, and Quality of Life of Older People
- Maternal psychiatric diagnosis and comorbidities around pregnancy
- Comparing the prevalence of multimorbidity using different operational definitions in primary care in Singapore based on a cross-sectional study using retrospective, large administrative data
- Sociodemographic and lifestyle predictors of incident hospital admissions with multimorbidity in a general population, 1999-2019: the EPIC-Norfolk cohort
- Cancer Survivorship Care Roles for Primary Care Physicians
- Development and validation of the Cambridge Multimorbidity Score
- Obesity is a public health emergency
- The Potpourri of Family Medicine, in Sickness and in Health