Patterns of Chronic Conditions and Their Association With Visual Impairment and Health Care Use | Geriatrics | JAMA Ophthalmology | JAMA Network
[Skip to Navigation]
Figure.  Probability of Having Chronic Conditions in Each Latent Class
Probability of Having Chronic Conditions in Each Latent Class
Table 1.  Probability of Reporting Chronic Conditions in Each Latent Class
Probability of Reporting Chronic Conditions in Each Latent Class
Table 2.  Distribution of Demographic and Health Behavior Variables by Latent Class Membershipa
Distribution of Demographic and Health Behavior Variables by Latent Class Membershipa
Table 3.  Odds Ratio of Reporting Visual Impairment for Different Chronic Condition Latent Classesa
Odds Ratio of Reporting Visual Impairment for Different Chronic Condition Latent Classesa
Table 4.  Previous 12-Month Health Care Utilization Comparison Between Chronic Condition Latent Class Groupsa
Previous 12-Month Health Care Utilization Comparison Between Chronic Condition Latent Class Groupsa
1.
Anderson  GF.  Chronic Care: Making the Case for Ongoing Care. Princeton, NJ: Robert Wood Johnson Foundation; 2010.
2.
Wolff  JL, Starfield  B, Anderson  G.  Prevalence, expenditures, and complications of multiple chronic conditions in the elderly.   Arch Intern Med. 2002;162(20):2269-2276. doi:10.1001/archinte.162.20.2269 PubMedGoogle ScholarCrossref
3.
Mollica  RL, Gillespie  J.  Care Coordination for People With Chronic Conditions. Portland, ME: Partnership for Solutions Portland; 2003.
4.
Gerteis  J, Izrael  D, Deitz  D,  et al.  Multiple Chronic Conditions Chartbook. Rockville, MD: Agency for Healthcare Research and Quality; 2014.
5.
Centers for Medicare & Medicaid Services.  Chronic Conditions Among Medicare Beneficiaries. Chartbook. 2012 ed. Baltimore, MD: US Dept of Health and Human Services; 2012.
6.
Schneider  KM, O’Donnell  BE, Dean  D.  Prevalence of multiple chronic conditions in the United States’ Medicare population.   Health Qual Life Outcomes. 2009;7(1):82. doi:10.1186/1477-7525-7-82 PubMedGoogle ScholarCrossref
7.
Goodman  RA, Posner  SF, Huang  ES, Parekh  AK, Koh  HK.  Defining and measuring chronic conditions: imperatives for research, policy, program, and practice.   Prev Chronic Dis. 2013;10:E66. doi:10.5888/pcd10.120239 PubMedGoogle Scholar
8.
Carabellese  C, Appollonio  I, Rozzini  R,  et al.  Sensory impairment and quality of life in a community elderly population.   J Am Geriatr Soc. 1993;41(4):401-407. doi:10.1111/j.1532-5415.1993.tb06948.x PubMedGoogle ScholarCrossref
9.
Bookwala  J, Lawson  B.  Poor vision, functioning, and depressive symptoms: a test of the activity restriction model.   Gerontologist. 2011;51(6):798-808. doi:10.1093/geront/gnr051 PubMedGoogle ScholarCrossref
10.
Wallhagen  MI, Strawbridge  WJ, Shema  SJ, Kurata  J, Kaplan  GA.  Comparative impact of hearing and vision impairment on subsequent functioning.   J Am Geriatr Soc. 2001;49(8):1086-1092. doi:10.1046/j.1532-5415.2001.49213.x PubMedGoogle ScholarCrossref
11.
Burmedi  D, Becker  S, Heyl  V, Wahl  H-W, Himmelsbach  I.  Emotional and social consequences of age-related low vision.   Vis Impair Res. 2002;4(1):47-71. doi:10.1076/vimr.4.1.47.15634 Google ScholarCrossref
12.
Pinquart  M, Pfeiffer  JP.  Psychological well-being in visually impaired and unimpaired individuals: a meta-analysis.   Br J Vis Impairment. 2011;29(1):27-45. doi:10.1177/0264619610389572 Google ScholarCrossref
13.
Klein  BE, Klein  R, Knudtson  MD, Lee  KE.  Relationship of measures of frailty to visual function: the Beaver Dam Eye Study.   Trans Am Ophthalmol Soc. 2003;101:191-196.PubMedGoogle Scholar
14.
Burmedi  D, Becker  S, Heyl  V, Wahl  H-W, Himmelsbach  I.  Behavioral consequences of age-related low vision.   Vis Impair Res. 2002;4(1):15-45. doi:10.1076/vimr.4.1.15.15633 Google ScholarCrossref
15.
Tinetti  ME, Allore  H, Araujo  KL, Seeman  T.  Modifiable impairments predict progressive disability among older persons.   J Aging Health. 2005;17(2):239-256. doi:10.1177/0898264305275176 PubMedGoogle ScholarCrossref
16.
Lam  BL, Christ  SL, Zheng  DD,  et al.  Longitudinal relationships among visual acuity and tasks of everyday life: the Salisbury Eye Evaluation Study.   Invest Ophthalmol Vis Sci. 2013;54(1):193-200. doi:10.1167/iovs.12-10542 PubMedGoogle ScholarCrossref
17.
Zheng  DD, Swenor  BK, Christ  SL, West  SK, Lam  BL, Lee  DJ.  Longitudinal associations between visual impairment and cognitive functioning: the Salisbury Eye Evaluation Study.   JAMA Ophthalmol. 2018;136(9):989-995. doi:10.1001/jamaophthalmol.2018.2493 PubMedGoogle ScholarCrossref
18.
Zhang  X, Gregg  EW, Cheng  YJ,  et al.  Diabetes mellitus and visual impairment: National Health and Nutrition Examination Survey, 1999-2004.   Arch Ophthalmol. 2008;126(10):1421-1427. doi:10.1001/archopht.126.10.1421 PubMedGoogle ScholarCrossref
19.
Jones  GC, Crews  JE, Danielson  ML.  Health risk profile for older adults with blindness: an application of the International Classification of Functioning, Disability, and Health framework.   Ophthalmic Epidemiol. 2010;17(6):400-410. doi:10.3109/09286586.2010.528137 PubMedGoogle ScholarCrossref
20.
Crews  JE, Campbell  VA.  Vision impairment and hearing loss among community-dwelling older Americans: implications for health and functioning.   Am J Public Health. 2004;94(5):823-829. doi:10.2105/AJPH.94.5.823 PubMedGoogle ScholarCrossref
21.
Verbrugge  LM, Jette  AM.  The disablement process.   Soc Sci Med. 1994;38(1):1-14. doi:10.1016/0277-9536(94)90294-1 PubMedGoogle ScholarCrossref
22.
Fried  TR, Tinetti  ME, Iannone  L.  Primary care clinicians’ experiences with treatment decision making for older persons with multiple conditions.   Arch Intern Med. 2011;171(1):75-80.PubMedGoogle Scholar
23.
Collins  LM, Lanza  ST.  Latent Class and Latent Transition Analysis: With Applications in the Social, Behavioral, and Health Sciences. New York, NY: John Wiley & Sons; 2010.
24.
Larsen  FB, Pedersen  MH, Friis  K, Glümer  C, Lasgaard  M.  A latent class analysis of multimorbidity and the relationship to socio-demographic factors and health-related quality of life: a national population-based study of 162,283 Danish adults.   PLoS One. 2017;12(1):e0169426. doi:10.1371/journal.pone.0169426 PubMedGoogle Scholar
25.
Whitson  HE, Johnson  KS, Sloane  R,  et al.  Identifying patterns of multimorbidity in older Americans: application of latent class analysis.   J Am Geriatr Soc. 2016;64(8):1668-1673. doi:10.1111/jgs.14201 PubMedGoogle ScholarCrossref
26.
Pugh  MJV, Finley  EP, Copeland  LA,  et al.  Complex comorbidity clusters in OEF/OIF veterans: the polytrauma clinical triad and beyond.   Med Care. 2014;52(2):172-181. doi:10.1097/MLR.0000000000000059 PubMedGoogle ScholarCrossref
27.
Centers for Disease Control and Prevention, National Center for Health Statistics. National Health Interview Survey: questionnaires, datasets, and related documentation. https://www.cdc.gov/nchs/nhis/nhis_questionnaires.htm. Accessed June 1, 2018.
28.
Wallace  R, Ackermann  R, Basen-Engquist  K,  et al.  Living well with chronic illness: a call for public health action.   Mil Med. 2015;180(5):485-487. doi:10.7205/MILMED-D-15-00034 PubMedGoogle ScholarCrossref
29.
National Academies of Sciences Engineering Medicine.  Hearing Health Care for Adults: Priorities for Improving Access and Affordability. Washington DC: The National Academies Press; 2016.
30.
Kessler  RC, Andrews  G, Colpe  LJ,  et al.  Short screening scales to monitor population prevalences and trends in non-specific psychological distress.   Psychol Med. 2002;32(6):959-976. doi:10.1017/S0033291702006074 PubMedGoogle ScholarCrossref
31.
Weissman  J, Pratt  LA, Miller  EA, Parker  JD.  Serious Psychological Distress Among Adults, United States, 2009-2013. Hyattsville, MD: National Center for Health Statistics; 2015. NCHS data brief 203.
32.
Nagin  DS.  Group-Based Modeling of Development. Cambridge, MA: Harvard University Press; 2005. doi:10.4159/9780674041318
33.
Parsons  VL, Moriarity  C, Jonas  K, Moore  TF, Davis  KE, Tompkins  L.  Design and estimation for the National Health Interview Survey, 2006-2015.   Vital Health Stat 2. 2014;(165):1-53.PubMedGoogle Scholar
34.
Asparouhov  T, Muthén  B.  Auxiliary variables in mixture modeling: three-step approaches using Mplus.   Struct Equ Modeling. 2014;21(3):329-341. doi:10.1080/10705511.2014.915181 Google ScholarCrossref
35.
Muthén  LK, Muthén  BO.  Mplus User’s Guide: Statistical Analysis With Latent Variables, User’s Guide. 8th ed. Los Angeles, CA: Muthén & Muthén; 2017.
36.
SAS Institute.  Base SAS 9.4 Procedures Guide: Statistical Procedures. 5th ed. Cary, NC: SAS Institute; 2017.
37.
Cattell  RB.  The scree test for the number of factors.   Multivariate Behav Res. 1966;1(2):245-276. doi:10.1207/s15327906mbr0102_10 PubMedGoogle ScholarCrossref
38.
Crews  JE, Chou  C-F, Sekar  S, Saaddine  JB.  The prevalence of chronic conditions and poor health among people with and without vision impairment, aged ≥65 years, 2010–2014.   Am J Ophthalmol. 2017;182:18-30. doi:10.1016/j.ajo.2017.06.038 PubMedGoogle ScholarCrossref
39.
Globe  DR, Varma  R, Torres  M, Wu  J, Klein  R, Azen  SP; Los Angeles Latino Eye Study Group.  Self-reported comorbidities and visual function in a population-based study: the Los Angeles Latino Eye Study.   Arch Ophthalmol. 2005;123(6):815-821. doi:10.1001/archopht.123.6.815 PubMedGoogle ScholarCrossref
40.
Li  Y, Crews  JE, Elam-Evans  LD,  et al.  Visual impairment and health-related quality of life among elderly adults with age-related eye diseases.   Qual Life Res. 2011;20(6):845-852. doi:10.1007/s11136-010-9825-z PubMedGoogle ScholarCrossref
41.
Brody  BL, Gamst  AC, Williams  RA,  et al.  Depression, visual acuity, comorbidity, and disability associated with age-related macular degeneration.   Ophthalmology. 2001;108(10):1893-1900. doi:10.1016/S0161-6420(01)00754-0 PubMedGoogle ScholarCrossref
42.
Fried  LP, Bandeen-Roche  K, Kasper  JD, Guralnik  JM.  Association of comorbidity with disability in older women: the Women’s Health and Aging Study.   J Clin Epidemiol. 1999;52(1):27-37. doi:10.1016/S0895-4356(98)00124-3 PubMedGoogle ScholarCrossref
43.
Verbrugge  LM, Lepkowski  JM, Imanaka  Y.  Comorbidity and its impact on disability.   Milbank Q. 1989;67(3-4):450-484. doi:10.2307/3350223 PubMedGoogle ScholarCrossref
44.
Islam  MM, Valderas  JM, Yen  L, Dawda  P, Jowsey  T, McRae  IS.  Multimorbidity and comorbidity of chronic diseases among the senior Australians: prevalence and patterns.   PLoS One. 2014;9(1):e83783. doi:10.1371/journal.pone.0083783 PubMedGoogle Scholar
45.
Ahmadi  S-F, Streja  E, Zahmatkesh  G,  et al.  Reverse epidemiology of traditional cardiovascular risk factors in the geriatric population.   J Am Med Dir Assoc. 2015;16(11):933-939. doi:10.1016/j.jamda.2015.07.014 PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    February 27, 2020

    Patterns of Chronic Conditions and Their Association With Visual Impairment and Health Care Use

    Author Affiliations
    • 1Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, Florida
    • 2Department of Human Development and Family Studies and Statistics, Purdue University, West Lafayette, Indiana
    • 3Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, Miami, Florida
    JAMA Ophthalmol. 2020;138(4):387-394. doi:10.1001/jamaophthalmol.2020.0052
    Key Points

    Question  What are the patterns of chronic conditions in the US population and their association with visual impairment and health care use?

    Findings  In this cross-sectional study of 13 years of National Health Interview Survey data for 387 780 individuals, 5 distinct groups of chronic condition patterns were identified: healthy, hypertensive, respiratory conditions, heart disease, and severely impaired. Individuals in each multimorbidity group showed distinct demographic and health behavior characteristics and different elevated risks of visual impairment and health care use.

    Meaning  The results may help clinicians, including ophthalmologists, to coordinate care and develop targeted interventions to prevent visual impairment and avert serious consequences of multiple chronic conditions.

    Abstract

    Importance  Visual impairment and visual disorders often co-occur with other chronic conditions. Understanding patterns of multimorbidity is important for reducing health care use and improving health outcomes.

    Objective  To identify chronic condition patterns and their association with visual impairment and health care use in a nationally representative sample.

    Design, Setting, and Participants  This cross-sectional study used National Health Interview Survey data for 387 780 individuals aged 18 years and older, representative of the civilian noninstitutionalized US population, from January 1, 2002, to December 31, 2014. Statistical analysis was performed from June to November 2018.

    Exposures  Participants were classified in subgroups with different combinations of self-reported chronic conditions using latent class analysis.

    Main Outcomes and Measures  Self-reported visual impairment, emergency department visit, and hospitalization use in the previous 12 months.

    Results  Among the 387 780 individuals included in the study, 51.8% were female, 77.6% were white, and the mean (SD) age was 46.2 (18.0) years. Latent class analysis identified 5 different classes, with 70.5% of the participants belonging to the healthy group. The other 4 groups represented various degrees and patterns of multimorbidity. The hypertensive group (19.6%) had a high prevalence of hypertension (62.6%), the respiratory conditions group (4.4%) had a high prevalence of emphysema (47.7%) and asthma (45.6%), the heart disease group (3.6%) had high prevalence of coronary heart disease (69.8%), and the severely impaired group (1.8%) had higher prevalence of most conditions compared with the other groups. In the adjusted analysis, compared with the healthy group, participants in all 4 disease groups had elevated risk of visual impairment: heart condition group (odds ratio [OR], 3.19; 95% CI, 2.92-3.48), hypertensive group (OR, 3.28; 95% CI, 3.10-3.48), respiratory condition group (OR, 3.87; 95% CI, 3.56-4.20), and severely impaired group (OR, 10.19; 95% CI, 9.20-11.28). All 4 disease groups had elevated risk of reporting emergency department use and hospitalization. For the severely impaired group, the OR for emergency department use was 9.39 (95% CI, 8.53-10.34), and the OR for hospitalization was 10.80 (95% CI, 9.80-11.92).

    Conclusions and Relevance  In this study, individuals in all 4 multimorbidity groups had an elevated risk of visual impairment and health care use compared with the healthy group. Characteristics of high-risk groups identified by this study may help in the development and implementation of interventions to avert the more serious consequences of having multiple chronic conditions.

    Introduction

    More people are living longer and developing chronic conditions. Multimorbidity is defined as the co-occurrence of at least 2 chronic conditions, and the number of older adults living with multimorbidity has increased rapidly.1 In a study of Medicare beneficiaries, 64% of participants had 2 or more conditions, and 24% had 4 or more conditions.2 Between 2000 and 2020, the number of Americans with multimorbidity is expected to increase from 60 million to 81 million.3 Approximately 71% of the total health care spending in the United States is for individuals with multiple chronic conditions.4 As the number of chronic conditions increases, associated medical costs increase.5,6 To address the increasing prevalence of multimorbidity, the US Department of Health and Human Services (DHHS) Initiative on Multiple Chronic Conditions established a standard list of conditions to guide research into multiple chronic conditions.7

    Visual impairment is among the most disabling chronic conditions and is associated with a broad range of psychosocial and functional consequences, including social isolation,8-10 depression,8-11 poor well-being,11,12 frailty,13 and functional and cognitive decline.14-17 For older persons, visual impairment often occurs with other disabling chronic conditions, including diabetes, hypertension, and respiratory problems.18-20 Multiple conditions combined with visual impairment influence daily functioning and social participation, and visual impairment may exacerbate the disabling effects of other chronic conditions.21

    Current health care, public health, and rehabilitation systems are largely designed to address individual conditions.22 Multimorbidity research has focused on the cumulative number of conditions rather than the qualitative combinations of conditions. Understanding the prevalent multimorbidity combinations is important in developing tailored interventions to treat the increasing number of people with multimorbidity. Latent class analysis (LCA) is a statistical approach that groups individuals into distinct homogeneous subgroups based on their pattern of response across several observed variables such that individuals within a group are more similar than individuals between groups.23

    Although LCA has been performed in the general population of other countries24 and in US subpopulations, such as older adults, to evaluate multimorbidity,25,26 to our knowledge, it has not been applied in studying the patterns of chronic conditions in the general US population, nor has the pattern of chronic conditions and its association with visual impairment been investigated. Our objective was to apply LCA to identify patterns of multimorbidity in a US nationally representative adult sample and to investigate the associations of the pattern of chronic conditions with visual impairment and health care use.

    Method
    Data Source

    For this cross-sectional study, we used the National Health Interview Survey (NHIS) pooled data (January 1, 2002, to December 31, 2014). The NHIS is a continuous multipurpose and multistage area probability survey of the civilian noninstitutionalized US population conducted by the National Center for Health Statistics.27 The NHIS is an annual, cross-sectional survey designed to provide nationally representative estimates on a wide range of health status and use measures. Each year, NHIS samples approximately 35 000 households including an estimated 87 500 persons, with an oversampling of black, Hispanic, and Asian race, plus adults aged 65 years or older.27 A core set of questions is asked each year, and data can be pooled across survey years to enhance sample size. The NHIS participants aged 18 years and older were included in this study. The University of Miami institutional review board, Miami, Florida, approved this study and exempted the patient consent form requirement because the study was a secondary analysis of publicly available deidentified data collected by the National Center for Health Statistics.

    Chronic Conditions

    The analysis included 13 chronic health conditions: hypertension, coronary heart disease, stroke, any heart condition (including angina, myocardial infarction, and other heart diseases), emphysema or chronic obstructive pulmonary disease (COPD), asthma, cancer, diabetes, arthritis, kidney disease, hepatitis, psychologic distress, and hearing impairment. Selected chronic conditions were based on the standard list of conditions established by a DHHS Initiative on Multiple Chronic Conditions7 that were available in NHIS. Participants were asked the following questions: “Have you ever been told by a doctor …that you had... (ie, hypertension, also called high blood pressure)?” We combined myocardial infarction, angina, and any other heart disease into a category called any heart condition.

    Hearing impairment was measured by the question: “Without hearing aids, is your hearing excellent, good, a little trouble hearing, moderate trouble, a lot of trouble, or are you deaf?” Answers other than excellent or good were considered as hearing impairment. Although the DHHS standard list does not include hearing impairment, we included it in our analysis because it is chronic,28,29 has a high prevalence, and limits activities of daily living.20 Nonspecific psychologic distress was measured by the Kessler-6 score (range, 0-24), with a sum score of 13 to 24 being psychologically distressed.30,31

    Outcome Variables

    Visual impairment was defined by the questions: “Do you have any trouble seeing even when wearing glasses or contact lenses?” A response of yes classified a participant as visually impaired.

    Emergency department (ED) visits were measured by the following: “During the past 12 months, how many times have you gone to a hospital emergency room about your own health?” Reporting 1 or more ED visits was coded as having an ED visit. Hospitalization was assessed by the following: “Were you in a hospital overnight in the past 12 months?”

    Statistical Analysis

    Statistical analysis was performed from June to November 2018. First, a series of LCA models ranging from 2 to 10 classes were estimated and compared. Model fit statistics used in comparing models included Akaike information criterion, bayesian information criterion, and sample size–adjusted bayesian information criterion, for which smaller values of fit indices indicate better model fit to the data. When a smallest value was not reached, a scree plot was used to select the optimal model. Entropy, an index that summarizes the overall precision of the classification for all samples across all latent classes, compared the LCA models. An entropy above 0.7 is considered a good value.32 The Vuong-Lo-Mendell-Rubin and Lo-Mendell-Rubin likelihood ratio tests were also used to evaluate model improvement as the number of classes increased. Also, the latent membership classes derived should have substantive clinical meaning and be distinct from each other. A parsimonious model is preferred when other indices are similar.

    Second, we explored the latent class membership associations with sociodemographic and health behavior variables using χ2 tests adjusted for the complex survey design of the NHIS data.33 The self-reported sociodemographic variables included age (years), sex (female or male), race (black, Asian, other races, or white), Hispanic background (Hispanic or non-Hispanic), educational level (high school graduate, more than a high school education, or less than a high school education), marital status (married or all other), health insurance (health insurance vs no health insurance), and household income status (income to poverty threshold ratio of <1.00, 1.00-1.99, >4.00, or unknown vs 2.00-3.99). The health behavior variables (categorical) included smoking status (current, former, or never), alcohol drinking status (none, moderate, heavy, or light), and body mass index (calculated as weight in kilograms divided by height in meters squared) (underweight [<18.5], overweight [25 to <30], obese [≥30], or normal weight [18.5 to <25]).

    We examined the association of the derived latent class membership with self-reported visual impairment status and indicators of health care use, including ED use and hospitalization. Logistic regression models with latent class membership as an independent variable and visual impairment and health care use as outcomes were performed. These models controlled for all aforementioned sociodemographic and health behavior variables. The regression model for the 2 health care use variables also controlled for visual impairment status. Introducing predictor or outcome variables to an LCA model could potentially cause latent class shifts; we adapted the manual 3-step approach34 to account for the measurement error associated while minimizing latent class shifts.

    All analyses accounted for the complex survey design with adjustments to the weight variable to account for the pooling of 13 years of data.27 The LCA and logistic regression analyses were performed using Mplus, version 8 (Muthén & Muthén).35 Other analyses were conducted using SAS, version 9.4 (SAS Institute Inc).36

    Results

    The total sample size was 387 780 individuals (51.8% female, 77.6% white, 11.8% black, 4.6% Asian, and 6% other race). The mean (SD) age was 46.2 (18.0) years. Hispanic individuals composed 13.6% of participants, and 56.8% had above a high school education. Of the 13 conditions included in the LCA, hypertension (28.1%) had the highest prevalence, followed by arthritis (22.1%) and hearing impairment (15.8%).

    Identifying Latent Class Chronic Disease Patterns in the Population

    The LCA model–fit statistics showed that as the number of classes increased, the value of the Akaike information criterion, bayesian information criterion, and sample size–adjusted bayesian information criterion decreased, and a smallest value was not reached (eTable 1 in the Supplement). Therefore, a scree plot37 was used to select the optimal model (eFigure in the Supplement). Minimum model improvement occurred beyond the 5-class model, which had the highest entropy value of 0.726. The 5-class model was selected based on a combination of model fit indices, parsimony, and clinical classification relevance. Participants were assigned to the class for which they had the highest probability of membership.

    Class proportions and the probabilities of having each chronic condition for each latent class are shown in Table 1. Classes were labeled according to their characteristic of the chronic condition probabilities (Figure): (1) the healthy group (70.5% of the sample) was characterized by low probabilities (<10.0%) of reporting any chronic condition; (2) the hypertensive group (19.6%) had a 62.6% probability of hypertension, 20.1% diabetes, and 48.0% arthritis; (3) the respiratory condition group (4.4%) was characterized by high prevalence of emphysema or COPD (47.7%) and asthma (45.6%); (4) the heart condition group (3.6%) had elevated coronary heart disease (69.8%), stroke (14.6%), and other heart conditions (85.2%); and (5) the severely impaired group (1.8%) showed the highest probabilities of reporting almost all chronic conditions (hypertension [88.9%], other heart condition [67.9%], arthritis [82.8%], emphysema or COPD [49.6%], diabetes [47.8%], and hearing impairment [54.3%]).

    Latent Class Membership, Socioeconomics, and Health Behaviors

    Table 2 depicts the characteristics of individuals in each class. Survey design–adjusted χ2 tests indicated that all socioeconomic and health behavior factors were distributed differently among classes. The healthy group and the respiratory condition group were generally younger (mean [SD] age, 40.2 [15.4] years and 52.1 [16.0] years), whereas the heart disease group and the severely impaired group were older (67.2 [12.4] years and 64.5 [12.1] years). The healthy group had the highest educational level (60.0% more than high school) and highest income (34.2% with income ratio ≥4), whereas the severely impaired group had the lowest educational level (36.9% with more than high school) and lowest income (13.2% with income ratio ≥4). The heart condition group had a high percentage of males (64.2%), whereas the respiratory condition group had a high percentage of females (66.6%). The heart conditions group had the highest proportion of married individuals (61.8%), whereas the severely impaired group had the lowest proportion (47.0%).

    The healthy group had the highest proportion of never smokers (63.5%), whereas the severely impaired group had the lowest (34.3%). In the respiratory condition group, 32.9% were current smokers. The healthy group had a high percentage of normal weight individuals (39.4%), and 49.9% of individuals in the severely impaired group were obese.

    Association of Latent Class Membership With Visual Impairment and Health Care Use

    Logistic regression indicated that all chronic condition latent classes had an elevated risk for reporting visual impairment (Table 3). Compared with the healthy group, the severely impaired group was the most likely to report visual impairment (odds ratio [OR], 10.19; 95% CI, 9.20-11.28), followed by the respiratory condition group (OR, 3.87; 95% CI, 3.56-4.20), the hypertensive group (OR, 3.28; 95% CI, 3.10-3.48), and the heart condition group (OR, 3.19; 95% CI, 2.92-3.48).

    Logistic regressions controlling for visual impairment and other covariates indicated that all chronic condition latent classes were more likely to report health care use than the healthy group (Table 4). Compared with the healthy group, the severely impaired group was most likely to report ED visits (OR, 9.39; 95% CI, 8.53-10.34) and hospitalization (OR, 10.80; 95% CI, 9.80-11.92), followed by the heart condition group (ED use: OR, 4.17 [95% CI, 3.90-4.46]; hospitalization: OR, 7.44 [95% CI, 6.89-8.03]), the respiratory condition group (ED use: OR, 3.67 [95% CI, 3.44-3.91]; hospitalization: OR, 3.08 [95% CI, 2.83-3.36]), and the hypertensive group (ED use: OR, 2.39 [95% CI, 2.29-2.49]; hospitalization: OR, 2.91 [95% CI, 2.74-3.08]) (2-sided P < .005 for all after Bonferroni correction for 10 comparisons [α = .05/10]).

    Discussion

    We defined patterns of chronic conditions in the US population and examined their association with self-reported visual impairment status and health care use using LCA techniques. Five distinct groups of chronic condition patterns were identified in the population: healthy, hypertensive, respiratory condition, heart condition, and severely impaired. Latent class analysis allowed us to reduce the complexity of chronic condition information in the population and characterize differences among the groups. We found that all chronic condition groups had elevated risk of reporting visual impairment and higher health care use compared with the healthy group.

    The 5 chronic condition groups identified by LCA represent different degrees of healthiness in the population, with the healthy group and the severely impaired group situated at the opposite ends of the spectrum and the other 3 groups each with its own characteristics in the middle. There appeared to be a direct association between the severity of the disease classes and the likelihood of reporting visual impairment and use of health care resources. The severely impaired group reported the highest health care use and the most visual impairment compared with all other groups.

    Studies that investigated the association between individual chronic conditions and visual impairment indicated a higher prevalence of chronic conditions among those with visual impairment compared with those without visual impairment.38 Such conditions include hypertension, diabetes, stroke, depression, chronic breathing problems, and hearing loss.18-20 Counts of chronic conditions have been shown to be higher among those living with visual impairment.39-41 Studies that assesses the presence of pairs of chronic conditions (dyads) found that visual impairment or vision disease was included in many of the leading pairs (ie, visual impairment and diabetes, visual impairment and stroke, visual impairment and hearing loss, and visual impairment and depression).42,43 In our study, visual impairment was associated with the patterns of chronic conditions. The severely impaired subpopulation, which had the poorest health, had the highest rate of reporting visual impairment. To our knowledge, this is the first time such an association between the patterns of chronic diseases and visual impairment has been reported using latent class modeling.

    Of interest, the heart condition group was similar in age (mean [SD], 67.2 [12.4] years) to the severely impaired group (mean [SD], 64.5 [12.1] years). However, the severely impaired group was less educated, had lower income, and was less likely to be married (Table 2). These socioeconomic factors may have contributed to worse health in the severely impaired group. Even though the severely impaired group was younger overall, these individuals were more likely to report ED use (OR, 2.25; 95% CI, 1.99-2.55) and more likely to be hospitalized (OR, 1.45; 95% CI, 1.28-1.64) compared with the heart condition group. The severely impaired group was also more likely to report visual impairment (OR, 3.19; 95% CI, 2.79-3.65). The socioeconomic disadvantage of the severely impaired group may have worked together with multiple chronic conditions to increase the likelihood of ED use and hospitalization. Socioeconomic disadvantage could be associated with less healthy lifestyles and less access to preventive or regular health care and thus overuse of the ED and hospitalization. These factors may lead to the development of more severe consequences of multiple chronic conditions in the severely impaired group.

    A previous study24 using LCA to analyze the Danish population was based on 15 chronic conditions and identified 7 multimorbidity classes. Three similar classes were identified in both studies. These include a healthy group composing most of the population, a hypertensive group, and a respiratory condition group. The Danish study24 identified complex cardiometabolic and complex respiratory classes, which were the 2 classes with the heaviest disease burden. This finding is similar to the 2 least healthy classes in our study, the heart condition group and the severely impaired group. The Danish study24 also included conditions that were not on the DHHS Initiative on Multiple Chronic Conditions list (back injuries, tinnitus, migraine, and cataract) or not available in NHIS (osteoporosis). In addition, we used the Kessler-6 index as a measure of psychologic distress, which could be different from the mental health disorder variables used in the Danish study.24 The population and health care systems in the United States and Denmark differ in many ways; therefore, study results are not generalizable to each other. Other studies used LCA to investigate disease patterns in subgroups of the population (eg, veterans).25,26,44 Results from these studies cannot be directly compared with the current study because of population differences.

    One study25 used Medicare beneficiary data to examine disease patterns of US adults aged 65 and older. This study concluded that health care use was higher in classes with greater morbidity, which is similar to our findings. However, this study concluded that individuals could not be assigned to a particular class with confidence because of a high misclassification error of 0.36. We also conducted a subgroup analysis using only individuals 65 years and older. Entropy, which measures the precision of the classification of LCA models, was substantially reduced to a range of 0.5 to 0.6 (eTable 2 in the Supplement) compared with values above 0.7 using the entire population. Such reduction in precision makes LCA less attractive when studying disease patterns in older adults possibly because of more heterogeneity among individuals within groups. This may be analogous to the reduced predictive utility of traditional risk factors in the older population.45

    Limitations

    The limitation of the NHIS is that all data are self-reported without objective confirmation. Another limitation of the study is that the list of chronic conditions was not comprehensive. Across NHIS years, there were only 13 of the 20 chronic conditions identified in the DHHS report7 available for analysis. Because the LCA method is data driven, the availability of the chronic condition data could affect resulting pattern identification. Nevertheless, the NHIS collects data from a representative sample of the US civilian population annually and is a powerful surveillance tool to examine chronic condition patterns in the US population.

    Conclusions

    This research contributes to a better understanding of the patterns of chronic diseases and health disparities in the US population. Our findings suggest that particular health conditions coexist and are associated with visual impairments. The results may provide insight for clinicians to enhance health care coordination. Results from this research may be valuable for identifying and targeting high-risk populations (eg, severely impaired individuals) who are vulnerable to health disparities and use a large amount of health care resources. Programs aiming to reduce health disparities could focus on subpopulations, such as those who are severely impaired. The characteristics of this group identified by this study may help others to develop and implement interventions that might avert the more serious consequences of multimorbidity through promoting preventive and regular care, managing patient complexity, and coordinating health care for populations at risk.

    Back to top
    Article Information

    Accepted for Publication: December 23, 2019.

    Corresponding Author: D. Diane Zheng, PhD, Department of Public Health Sciences, University of Miami Miller School of Medicine, 1120 NW 14th St, Clinical Research Building–1070A, Miami FL 33136 (dzheng@med.miami.edu).

    Published Online: February 27, 2020. doi:10.1001/jamaophthalmol.2020.0052

    Author Contributions: Dr Zheng had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Zheng, Christ, McCollister, Lee.

    Acquisition, analysis, or interpretation of data: Zheng, Christ, Lam, Feaster, Lee.

    Drafting of the manuscript: Zheng, Lee.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Zheng, Christ, Feaster.

    Obtained funding: Zheng.

    Administrative, technical, or material support: Lam.

    Supervision: Christ, McCollister.

    Conflict of Interest Disclosures: None reported.

    Funding/Support: This work was supported by grant NEI F31EY025936 from the National Eye Institute.

    Role of the Funder/Sponsor: The funding organization had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Meeting Presentation: This study was presented in part at the 2019 Association for Research in Vision Ophthalmology Conference; April 29, 2019; Vancouver, British Columbia, Canada.

    References
    1.
    Anderson  GF.  Chronic Care: Making the Case for Ongoing Care. Princeton, NJ: Robert Wood Johnson Foundation; 2010.
    2.
    Wolff  JL, Starfield  B, Anderson  G.  Prevalence, expenditures, and complications of multiple chronic conditions in the elderly.   Arch Intern Med. 2002;162(20):2269-2276. doi:10.1001/archinte.162.20.2269 PubMedGoogle ScholarCrossref
    3.
    Mollica  RL, Gillespie  J.  Care Coordination for People With Chronic Conditions. Portland, ME: Partnership for Solutions Portland; 2003.
    4.
    Gerteis  J, Izrael  D, Deitz  D,  et al.  Multiple Chronic Conditions Chartbook. Rockville, MD: Agency for Healthcare Research and Quality; 2014.
    5.
    Centers for Medicare & Medicaid Services.  Chronic Conditions Among Medicare Beneficiaries. Chartbook. 2012 ed. Baltimore, MD: US Dept of Health and Human Services; 2012.
    6.
    Schneider  KM, O’Donnell  BE, Dean  D.  Prevalence of multiple chronic conditions in the United States’ Medicare population.   Health Qual Life Outcomes. 2009;7(1):82. doi:10.1186/1477-7525-7-82 PubMedGoogle ScholarCrossref
    7.
    Goodman  RA, Posner  SF, Huang  ES, Parekh  AK, Koh  HK.  Defining and measuring chronic conditions: imperatives for research, policy, program, and practice.   Prev Chronic Dis. 2013;10:E66. doi:10.5888/pcd10.120239 PubMedGoogle Scholar
    8.
    Carabellese  C, Appollonio  I, Rozzini  R,  et al.  Sensory impairment and quality of life in a community elderly population.   J Am Geriatr Soc. 1993;41(4):401-407. doi:10.1111/j.1532-5415.1993.tb06948.x PubMedGoogle ScholarCrossref
    9.
    Bookwala  J, Lawson  B.  Poor vision, functioning, and depressive symptoms: a test of the activity restriction model.   Gerontologist. 2011;51(6):798-808. doi:10.1093/geront/gnr051 PubMedGoogle ScholarCrossref
    10.
    Wallhagen  MI, Strawbridge  WJ, Shema  SJ, Kurata  J, Kaplan  GA.  Comparative impact of hearing and vision impairment on subsequent functioning.   J Am Geriatr Soc. 2001;49(8):1086-1092. doi:10.1046/j.1532-5415.2001.49213.x PubMedGoogle ScholarCrossref
    11.
    Burmedi  D, Becker  S, Heyl  V, Wahl  H-W, Himmelsbach  I.  Emotional and social consequences of age-related low vision.   Vis Impair Res. 2002;4(1):47-71. doi:10.1076/vimr.4.1.47.15634 Google ScholarCrossref
    12.
    Pinquart  M, Pfeiffer  JP.  Psychological well-being in visually impaired and unimpaired individuals: a meta-analysis.   Br J Vis Impairment. 2011;29(1):27-45. doi:10.1177/0264619610389572 Google ScholarCrossref
    13.
    Klein  BE, Klein  R, Knudtson  MD, Lee  KE.  Relationship of measures of frailty to visual function: the Beaver Dam Eye Study.   Trans Am Ophthalmol Soc. 2003;101:191-196.PubMedGoogle Scholar
    14.
    Burmedi  D, Becker  S, Heyl  V, Wahl  H-W, Himmelsbach  I.  Behavioral consequences of age-related low vision.   Vis Impair Res. 2002;4(1):15-45. doi:10.1076/vimr.4.1.15.15633 Google ScholarCrossref
    15.
    Tinetti  ME, Allore  H, Araujo  KL, Seeman  T.  Modifiable impairments predict progressive disability among older persons.   J Aging Health. 2005;17(2):239-256. doi:10.1177/0898264305275176 PubMedGoogle ScholarCrossref
    16.
    Lam  BL, Christ  SL, Zheng  DD,  et al.  Longitudinal relationships among visual acuity and tasks of everyday life: the Salisbury Eye Evaluation Study.   Invest Ophthalmol Vis Sci. 2013;54(1):193-200. doi:10.1167/iovs.12-10542 PubMedGoogle ScholarCrossref
    17.
    Zheng  DD, Swenor  BK, Christ  SL, West  SK, Lam  BL, Lee  DJ.  Longitudinal associations between visual impairment and cognitive functioning: the Salisbury Eye Evaluation Study.   JAMA Ophthalmol. 2018;136(9):989-995. doi:10.1001/jamaophthalmol.2018.2493 PubMedGoogle ScholarCrossref
    18.
    Zhang  X, Gregg  EW, Cheng  YJ,  et al.  Diabetes mellitus and visual impairment: National Health and Nutrition Examination Survey, 1999-2004.   Arch Ophthalmol. 2008;126(10):1421-1427. doi:10.1001/archopht.126.10.1421 PubMedGoogle ScholarCrossref
    19.
    Jones  GC, Crews  JE, Danielson  ML.  Health risk profile for older adults with blindness: an application of the International Classification of Functioning, Disability, and Health framework.   Ophthalmic Epidemiol. 2010;17(6):400-410. doi:10.3109/09286586.2010.528137 PubMedGoogle ScholarCrossref
    20.
    Crews  JE, Campbell  VA.  Vision impairment and hearing loss among community-dwelling older Americans: implications for health and functioning.   Am J Public Health. 2004;94(5):823-829. doi:10.2105/AJPH.94.5.823 PubMedGoogle ScholarCrossref
    21.
    Verbrugge  LM, Jette  AM.  The disablement process.   Soc Sci Med. 1994;38(1):1-14. doi:10.1016/0277-9536(94)90294-1 PubMedGoogle ScholarCrossref
    22.
    Fried  TR, Tinetti  ME, Iannone  L.  Primary care clinicians’ experiences with treatment decision making for older persons with multiple conditions.   Arch Intern Med. 2011;171(1):75-80.PubMedGoogle Scholar
    23.
    Collins  LM, Lanza  ST.  Latent Class and Latent Transition Analysis: With Applications in the Social, Behavioral, and Health Sciences. New York, NY: John Wiley & Sons; 2010.
    24.
    Larsen  FB, Pedersen  MH, Friis  K, Glümer  C, Lasgaard  M.  A latent class analysis of multimorbidity and the relationship to socio-demographic factors and health-related quality of life: a national population-based study of 162,283 Danish adults.   PLoS One. 2017;12(1):e0169426. doi:10.1371/journal.pone.0169426 PubMedGoogle Scholar
    25.
    Whitson  HE, Johnson  KS, Sloane  R,  et al.  Identifying patterns of multimorbidity in older Americans: application of latent class analysis.   J Am Geriatr Soc. 2016;64(8):1668-1673. doi:10.1111/jgs.14201 PubMedGoogle ScholarCrossref
    26.
    Pugh  MJV, Finley  EP, Copeland  LA,  et al.  Complex comorbidity clusters in OEF/OIF veterans: the polytrauma clinical triad and beyond.   Med Care. 2014;52(2):172-181. doi:10.1097/MLR.0000000000000059 PubMedGoogle ScholarCrossref
    27.
    Centers for Disease Control and Prevention, National Center for Health Statistics. National Health Interview Survey: questionnaires, datasets, and related documentation. https://www.cdc.gov/nchs/nhis/nhis_questionnaires.htm. Accessed June 1, 2018.
    28.
    Wallace  R, Ackermann  R, Basen-Engquist  K,  et al.  Living well with chronic illness: a call for public health action.   Mil Med. 2015;180(5):485-487. doi:10.7205/MILMED-D-15-00034 PubMedGoogle ScholarCrossref
    29.
    National Academies of Sciences Engineering Medicine.  Hearing Health Care for Adults: Priorities for Improving Access and Affordability. Washington DC: The National Academies Press; 2016.
    30.
    Kessler  RC, Andrews  G, Colpe  LJ,  et al.  Short screening scales to monitor population prevalences and trends in non-specific psychological distress.   Psychol Med. 2002;32(6):959-976. doi:10.1017/S0033291702006074 PubMedGoogle ScholarCrossref
    31.
    Weissman  J, Pratt  LA, Miller  EA, Parker  JD.  Serious Psychological Distress Among Adults, United States, 2009-2013. Hyattsville, MD: National Center for Health Statistics; 2015. NCHS data brief 203.
    32.
    Nagin  DS.  Group-Based Modeling of Development. Cambridge, MA: Harvard University Press; 2005. doi:10.4159/9780674041318
    33.
    Parsons  VL, Moriarity  C, Jonas  K, Moore  TF, Davis  KE, Tompkins  L.  Design and estimation for the National Health Interview Survey, 2006-2015.   Vital Health Stat 2. 2014;(165):1-53.PubMedGoogle Scholar
    34.
    Asparouhov  T, Muthén  B.  Auxiliary variables in mixture modeling: three-step approaches using Mplus.   Struct Equ Modeling. 2014;21(3):329-341. doi:10.1080/10705511.2014.915181 Google ScholarCrossref
    35.
    Muthén  LK, Muthén  BO.  Mplus User’s Guide: Statistical Analysis With Latent Variables, User’s Guide. 8th ed. Los Angeles, CA: Muthén & Muthén; 2017.
    36.
    SAS Institute.  Base SAS 9.4 Procedures Guide: Statistical Procedures. 5th ed. Cary, NC: SAS Institute; 2017.
    37.
    Cattell  RB.  The scree test for the number of factors.   Multivariate Behav Res. 1966;1(2):245-276. doi:10.1207/s15327906mbr0102_10 PubMedGoogle ScholarCrossref
    38.
    Crews  JE, Chou  C-F, Sekar  S, Saaddine  JB.  The prevalence of chronic conditions and poor health among people with and without vision impairment, aged ≥65 years, 2010–2014.   Am J Ophthalmol. 2017;182:18-30. doi:10.1016/j.ajo.2017.06.038 PubMedGoogle ScholarCrossref
    39.
    Globe  DR, Varma  R, Torres  M, Wu  J, Klein  R, Azen  SP; Los Angeles Latino Eye Study Group.  Self-reported comorbidities and visual function in a population-based study: the Los Angeles Latino Eye Study.   Arch Ophthalmol. 2005;123(6):815-821. doi:10.1001/archopht.123.6.815 PubMedGoogle ScholarCrossref
    40.
    Li  Y, Crews  JE, Elam-Evans  LD,  et al.  Visual impairment and health-related quality of life among elderly adults with age-related eye diseases.   Qual Life Res. 2011;20(6):845-852. doi:10.1007/s11136-010-9825-z PubMedGoogle ScholarCrossref
    41.
    Brody  BL, Gamst  AC, Williams  RA,  et al.  Depression, visual acuity, comorbidity, and disability associated with age-related macular degeneration.   Ophthalmology. 2001;108(10):1893-1900. doi:10.1016/S0161-6420(01)00754-0 PubMedGoogle ScholarCrossref
    42.
    Fried  LP, Bandeen-Roche  K, Kasper  JD, Guralnik  JM.  Association of comorbidity with disability in older women: the Women’s Health and Aging Study.   J Clin Epidemiol. 1999;52(1):27-37. doi:10.1016/S0895-4356(98)00124-3 PubMedGoogle ScholarCrossref
    43.
    Verbrugge  LM, Lepkowski  JM, Imanaka  Y.  Comorbidity and its impact on disability.   Milbank Q. 1989;67(3-4):450-484. doi:10.2307/3350223 PubMedGoogle ScholarCrossref
    44.
    Islam  MM, Valderas  JM, Yen  L, Dawda  P, Jowsey  T, McRae  IS.  Multimorbidity and comorbidity of chronic diseases among the senior Australians: prevalence and patterns.   PLoS One. 2014;9(1):e83783. doi:10.1371/journal.pone.0083783 PubMedGoogle Scholar
    45.
    Ahmadi  S-F, Streja  E, Zahmatkesh  G,  et al.  Reverse epidemiology of traditional cardiovascular risk factors in the geriatric population.   J Am Med Dir Assoc. 2015;16(11):933-939. doi:10.1016/j.jamda.2015.07.014 PubMedGoogle ScholarCrossref
    ×