Health of Community-Dwelling Adults With Mobility Limitations in the United States: Incidence of Secondary Health Conditions. Part II

Data Source 

The 1996−1997 Medical Expenditure Panel Survey (MEPS)49 and the 1995 National Health Interview Survey (NHIS) Disability Supplement50 provided data for this study. The 1995 NHIS was the sampling frame for the 1996 MEPS, using a multistage probability sampling design. Thus, the 1996 MEPS included a subsample of NHIS households that were nationally representative of the civilian, noninstitutionalized U.S. population, which allowed data from the same respondents from both surveys to be used. In 1995, a disability supplement to the NHIS (NHIS-D) included data on self-reported limitations and use of assistive equipment that was not available from the MEPS, which augmented limitation group classification. The MEPS used an overlapping panel design such that 2 panels of respondents are included in the data set in any one year with the exception of 1996, the first year that the MEPS was fielded. Over a 2-year period, data were collected longitudinally through 5 rounds of household interviews. Only respondents in panel 1 were included in this analysis, beginning with round 1 of 1996 through round 5 of 1997. The primary advantages of linking the 1995 NHIS with the 1996−1997 MEPS were that (1) the 1994−1995 NHIS included unique indicators of disability from the disability supplement, which has not been repeated since that time; and (2) respondents with disabilities were oversampled in the 1996−1997 MEPS (panel 1). Oversampling of adults with disabilities provided a sufficiently large sample to address our research questions. Survey design and methods for the NHIS,51, 52 the NIHS-D,53 and the MEPS54, 55, 56 have been described.

Respondents 

Starting with all respondents on the 1996 MEPS full year consolidated file (study population), the analytic sample was limited in a previous, related study57 to 13,979 adults (18 years of age and older) with both non-zero (and therefore useable) analytic weights necessary for producing population estimates and data on the NHIS-D, which contained selected variables necessary for limitation group classification. The analytic sample was further limited for the current study. Of the 13,979 adults, 12,317 were in-scope (that is, part of the civilian, noninstitutionalized U.S. population) for all 5 rounds of data collection. Fifteen had insufficient data for limitation group classification. Thus, the analytic sample comprised 12,302 adults (fig 1). The institutional review board (IRB) of the University of Maryland, Baltimore determined this project to be exempt from the IRB approval process (exemption no. MR-060301).

View Large Image Download to PowerPoint

Fig 1. Schematic diagram of persons included and excluded from the final analytic sample, beginning with all 1996 MEPS adults with non-zero analytic weights that had data from the 1995 NHIS-D. White boxes depict persons excluded from the analytic sample.

Analytic Variables 

Respondents were classified into 3 groups for analysis: those with mobility limitations, nonmobility limitations, and no limitations. Variables from the 1996 MEPS, round 1, panel 1, and the NHIS-D were used primarily for limitation group classification.57 Adults reporting difficulty climbing stairs, walking, standing, or bending/stooping, or who reported use of mobility devices were classified with mobility limitation. Adults without mobility limitations who reported any other types of limitations (nonmobility) or who reported use of assistive technology other than mobility devices were classified as having other limitations. All other adults were classified as having no limitations. Methods for limitation group classification have been described in more detail elsewhere.57

Self-reported health conditions (open-ended account of physical conditions, injuries, and mental or emotional health conditions) were enumerated at the beginning of each round of data collection spaced approximately 4 months apart. Respondents were asked to include all of their conditions regardless of whether they saw a medical provider, received treatment, or took medications. Conditions associated with health care provider visits, hospitalizations, prescription medication use, and disability days were also reported by respondents in other sections of the MEPS. Interviewers recorded conditions as verbatim text, which were assigned 1996 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes by professional coders. ICD-9-CM codes were aggregated into 259 mutually exclusive, clinically homogeneous categories using Clinical Classification Software (CCS) developed by the Agency for Healthcare Research and Quality (AHRQ).58, 59, 60 CCS categories were used for this analysis. Prevalent conditions were identified in round 1 and differentiated from incident conditions occurring in rounds 2 through 5 using a flag that labeled the round in which the condition was first reported. The numerator for estimates of the percentage of adults with incident conditions comprised adults with newly occurring conditions (ie, those that occurred in rounds 2 through 5 but were not present in round 1). The denominator included those at risk for acquiring the condition in rounds 2 through 5 (ie, those that did not have the condition in round 1). In addition, men were removed from the denominators of conditions that only women could acquire (eg, cancer of the uterus, or ovarian cysts), and women were removed from the denominators of conditions that only men could acquire (eg, cancer of the prostate or other male genital disorders). CCS condition categories were used to define the dependent variables in analyses of incident conditions. Incident conditions were distinguished from prevalent conditions if they fell into different CCS condition categories. Thus, if a respondent reported a condition in CCS category 204 (other nontraumatic joint disorders) at round 1 and reported a different condition within that category at round 2, it was not counted as incident. Cumulative incidences were reported for rounds 2 through 5, covering roughly a 2-year period from 1996 to 1997.

Individual descriptors included age, sex, race and ethnicity, and body mass index (BMI). Marital status, household size, education, income (1996 family income as a percentage of the national poverty threshold), health insurance coverage (summary indicator of coverage for persons in the 1996 calendar year), and access to health care (only collected in round 2) were conceptualized as potential resources. Individual descriptors and resources reported at round 1 were used for analysis except where otherwise indicated. Variable construction has been described elsewhere.57

Statistical Analyses 

We explored potential bias due to exclusion of adults from the analytical sample because of withdrawal (death, institutionalization, military service, relocation outside of the United States, or nonfielding) or loss to follow-up by: (1) comparing selected limitation-related factors, individual characteristics, availability of resources, and health status of adults who were in-scope for all 5 rounds of data collection versus those who withdrew or were lost to follow-up; (2) calculating the mean number and 95% confidence intervals (CIs) of prevalent conditions at round 1 for adults who were in-scope, withdrew, or were lost to follow-up; and (3) calculating the mean number of incident conditions for 1996 for in-scope adults and those lost to follow-up (because data were available for the full year in these 2 groups). MEPS 1996 sample weights were used to produce population estimates for these comparisons by adjusting for differential selection probability.

Two-year incidence estimates of the most frequently occurring conditions (of 259 CCS categories) were rank-ordered for each limitation group. CCS categories were then collapsed into system-level categories and rank ordered. Because many of the high-ranking CCS categories comprised “other” conditions (ie, residual categories) and were therefore nonspecific, the most frequently occurring ICD-9-CM code that reflected a specific condition within each of these categories was determined to enhance clinical meaningfulness. However, there were instances when all of the most frequently occurring ICD-9 codes within a CCS category reflected a residual category rather than a specific condition. In these instances, estimates for individual ICD-9 codes were not reported. Respondents reporting more than 1 condition within a system-level category were counted only once in that category for the purpose of generating system-level incidence estimates. System-level categorization developed at AHRQ60 was modified as previously described57 such that sensory conditions were separated from “diseases of the nervous system,” and events such as immunizations were separated from “other conditions” and labeled as “health services encounters.” Presence or absence of incident conditions based on the CCS and system level condition categories were the dependent variables (outcomes) in logistic regression analyses examining differences in incident conditions across limitation groups while controlling for individual characteristics (sex, age, race and ethnicity, obesity) and availability of resources (marital status, household size, education, income, health insurance, access to health care). Differences in the number of incident conditions across limitation groups were examined with a linear regression model controlling for the number of prevalent conditions in addition to the factors noted above. Potentially significant and meaningful 2-way interactions between limitation group and all other independent variables were explored for possible inclusion in the final linear regression model. In all analyses of incident conditions, MEPS longitudinal sample weights for 1996−1997 were used to produce population estimates by adjusting for differential selection probability, withdrawal, and loss to follow-up.

Variance estimation was accomplished through Taylor linearization taking into account the complex sampling design. All estimates were based on a minimum of 30 responses per analytic cell with relative standard errors (SEs) less than 30%.53 The threshold for statistical significance was P less than .05. SASa and SUDAANb software programs were used for statistical analyses. SEs are reported in parentheses unless otherwise indicated.

Analytic Sample 

The follow-up status of 1996 MEPS adults was compared according to selected limitation-related factors, individual descriptors, resources, and health status (table 1). Of the 13,979 adults who had data from the 1995 NHIS-D and useable (non-zero) analytic weights, 12,317 or 87.2% (weighted estimate) were in-scope for all 5 rounds of data collection (ie, part of the civilian, noninstitutionalized U.S. population), 2.8% withdrew, and 10.0% were lost to follow-up. The mean number of prevalent conditions at round 1 (95% CIs) was greater for adults who withdrew (mean, 2.8; 95% CI, 2.5−3.0) compared with in-scope adults (mean, 1.7; 95% CI, 1.6−1.7) and those who were lost to follow-up (mean, 1.5; 95% CI, 1.4−1.6).

Table 1. Follow-Up Status of 1996 MEPS Adults Described by Selected Limitation-Related Factors, Demographic Characteristics, Resources, and Health Status

	Descriptive Variables	Percent ± SE⁎			P (chi-square)
		In-scope†	Withdrew	Lost
	Limitation group‡				<.001
	Mobility limitation	81.8±1.2	10.5±1.0	7.7±0.8
	Nonmobility limitation	89.4±1.2	4.5±0.8	6.2±1.0
	No limitation	88.2±0.6	1.0±0.1	10.8±0.6
	Age (y)				<.001
	18−44	88.1±0.6	0.9±0.1	10.9±0.6
	45−64	88.2±0.8	1.3±0.2	10.5±0.8
	≥65	82.7±1.0	11.0±0.7	6.3±0.7
	Sex				.002
	Men	86.3±0.6	3.2±0.2	10.5±0.6
	Women	88.1±0.5	2.4±0.2	9.6±0.5
	Race and ethnicity				<.001
	Non-Hispanic white	86.6±0.6	2.7±0.2	10.7±0.6
	Non-Hispanic black	87.1±1.2	3.6±0.6	9.3±1.0
	Hispanic	91.4±0.7	2.4±0.4	6.2±0.7
	Other	90.1±1.6	2.0±0.7	7.9±1.6
	Obesity (kg/m2)‡				.099
	BMI ≥30	88.7±0.9	2.2±0.4	9.0±0.8
	BMI <30	87.0±0.5	2.9±0.2	10.1±0.5
	Marital status‡				<.001
	Spouse in the house	87.3±0.7	2.0±0.2	10.6±0.7
	Not married/no spouse	87.1±0.6	3.7±0.3	9.2±0.6
	Household size				<.001
	Living alone	88.8±0.7	5.0±0.5	6.3±0.6
	2	86.2±0.8	3.1±0.3	10.7±0.8
	3−4	87.4±0.9	1.7±0.2	10.9±0.8
	≥5	87.2±1.4	1.7±0.3	11.1±1.4
	Education‡				<.001
	< High school	86.5±0.8	6.3±0.5	7.2±0.6
	High school	87.7±0.7	2.2±0.2	10.1±0.7
	> High school	87.4±0.6	1.4±0.2	11.3±0.6
	Poverty status				<.001
	Poor (<100% poverty line§)	86.4±1.0	7.2±0.7	6.4±0.8
	Low income (100%−199% poverty line)	88.3±0.8	4.5±0.5	7.3±0.7
	Middle income (200%—399% poverty line)	87.5±0.8	2.0±0.2	10.6±0.8
	High income (≥400% poverty line)	86.7±0.8	1.2±0.2	12.1±0.8
	Health status‡				<.001
	Excellent/very good	88.0±0.6	1.2±0.1	10.8±0.6
	Good	88.2±0.7	2.7±0.4	9.1±0.7
	Fair/poor	83.9±1.0	7.8±0.8	8.3±0.8

⁎ Proportions may not add to 100% due to rounding. † Respondent was considered in-scope if he/she was a member of the U.S. civilian, noninstitutionalized population during all 5 rounds of the 1996-1997 MEPS. ‡ Missing values excluded from estimates (≤0.4% for limitation group, 4.0% for obesity, .01% for marital status, 0.8% for education, 0.4% for health status). § Poverty line: 1996 family income as percentage of 1996 national poverty threshold determined by the U.S. Census Bureau.

Reasons for withdrawal included death, institutionalization, military service, and relocation outside of the United States. Of the 13,979 adults, 1.8% died over the 2-year study period, 0.7% were institutionalized, and 0.3% withdrew for other reasons. By far, the majority of withdrawals were due to death or institutionalization. There were more deaths and institutionalizations (weighted estimate ± SE) among adults with mobility limitations (10.4%±1.0%) compared with those with nonmobility limitations (4.5%±0.8%) or those without limitations (0.7%±0.1%). As shown in table 1, withdrawals (deaths/institutionalizations) were greater among adults who were 65 years of age or older, men, non-Hispanic black, not married, living alone, with less than a high school education, in poverty, and in fair to poor health.

In contrast to adults who died or were institutionalized, more of the adults who were lost to follow-up did not report limitations, were young (18−44 years old), men, non-Hispanic white, and married with multiple family members. They were also more likely to have more than a high school education, a high income, and excellent to very good health. The mean number of incident conditions for 1996 was greater for in-scope adults (mean, 1.8; 95% CI, 1.7−1.8) compared with those who were lost to follow-up (mean, 1.4; 95% CI, 1.3−1.6).

Description of Limitation Groups 

Limitation group differences described previously57 are summarized here. A greater proportion of adults with mobility limitations were 65 years of age or older (43% of this analytic sample) compared with those with nonmobility limitations (33%) or without limitations (10%), and more were women and obese. In comparison with other community-dwelling adults, a greater proportion of adults with mobility limitations were also unmarried, lived alone, had less than a high school education, were low-income or poor, and had only public insurance.

Incident Health Conditions 

Incidence rates for the majority of conditions were much higher in adults with mobility limitations compared with those without limitations (table 2). There was generally a gradient in the magnitude of incidence such that adults with mobility limitations had the highest incidences, followed by adults with nonmobility limitations. Adults without limitations generally had the lowest incidences.

Table 2. Two-Year Cumulative Incidence (U.S. 1996−1997) and Rank of the 25 Health Conditions (out of 259 condition categories)59, 60 Most Frequently Self-Reported by Adults With Mobility Limitations Compared to Those Same Conditions in the Remaining 2 Groups (incident conditions also shown by system-level categories)

	Condition Category	2-Year Incidence and Frequency Rank (Percent ± SE)
		Mobility Limitation Incidence	Nonmobility Limitation Incidence	No Limitation Incidence	Mobility Limitation Rank	Nonmobility Limitation Rank	No Limitation Rank
	CCS condition categories59⁎
	Other upper respiratory infections59†	24.6±1.5	24.7±1.8	25.7±0.6	1	1	1
	Acute nasopharyngitis (common cold)‡	14.4±1.2	15.7±1.5	15.9±0.5
	Other nontraumatic joint disorders59§	21.9±1.5	12.0±1.2	6.1±0.3	2	3	10
	Other and unspecified arthropathies‡	13.7±1.3	6.8±0.9	2.6±0.2
	Residual codes, unclassified59∥	16.5±1.2	11.4±1.2	5.4±0.3	3	6	12
	General symptoms‡	8.4±0.9	6.3±0.9	2.2±0.2
	Intestinal infection	14.9±1.2	15.2±1.5	16.0±0.5	4	2	2
	Other connective tissue disease59¶	14.6±1.0	11.5±1.1	6.3±0.3	5	4	9
	Peripheral enthesopathies and allied syndromes‡	2.6±0.5	2.3±0.5	1.7±0.2
	Spondylosis, intervertebral disk disorders, other back problems	14.1±1.1	11.5±1.2	7.7±0.3	6	5	4
	Other skin disorders59#	12.3±0.9	8.6±1.1	7.1±0.3	7	9	5
	Symptoms involving skin and other integumentary tissue (such as disturbance of sensation)‡, ###	4.4±0.6	1.9±0.5	2.3±0.2
	Essential hypertension	11.6±1.1	8.4±1.1	4.2±0.2	8	11	19
	Other upper-respiratory disease59⁎⁎	11.5±1.0	10.6±1.4	9.0±0.4	9	7	3
	Allergic rhinitis‡	6.4±0.8	5.4±1.0	6.1±0.3
	Other mental conditions59††	10.8±1.1	7.5±1.1	4.2±0.3	10	13	18
	Depressive disorder, not elsewhere classified‡	9.5±1.0	6.4±1.0	3.7±0.3
	Other lower-respiratory disease59‡‡	10.3±1.0	4.9±0.8	2.9±0.2	11	27	33
	Symptoms involving respiratory system and other chest symptoms (eg, dyspnea and chest pain)‡	6.4±0.8	3.6±0.7	1.8±0.2
	Urinary tract infections	9.2±1.0	6.8±0.9	4.5±0.2	13	18	17
	Other injuries and conditions due to external causes59§§	9.0±0.9	7.3±1.0	5.2±0.3	14	14	14
	Disorders of teeth and jaw	8.6±0.9	6.3±0.9	6.6±0.3	15	19	8
	Other disorders of stomach and duodenum59∥∥	8.2±0.9	6.9±0.9	3.5±0.2	16	17	25
	Disorders of function of stomach (such as gastroparesis or dyspepsia)‡, ⁎⁎⁎⁎	6.4±0.8	5.4±0.8	3.0±0.2
	COPD and bronchiectasis	7.7±0.7	7.7±1.2	4.8±0.3	17	12	15
	Other gastrointestinal disorders59¶¶	7.5±0.7	6.0±0.9	2.6±0.2	18	21	36
	Symptoms involving digestive system (such as dysphagia)‡, ††††	4.3±0.5	3.6±0.7	1.6±0.1
	Headache, including migraine	7.1±0.9	4.9±0.8	5.4±0.3	20	28	13
	Sprains and strains	6.9±0.8	7.2±0.9	7.1±0.3	21	16	6
	Other eye disorders59##	6.7±0.8	5.0±0.8	3.0±0.2	22	26	29
	Influenza	6.6±0.7	6.0±0.9	6.0±0.3	23	22	11
	Other nervous system disorders59⁎⁎⁎	6.5±0.7	3.8±0.7	2.2±0.2	24	37	39
	Mononeuritis of upper limb and mononeuritis multiplex‡	2.2±0.4	‡‡‡‡	0.7±0.1
	Fluid and electrolyte disorders	6.0±0.8	3.1±0.7	1.3±0.1	25	45	55
	Other ear and sense organ disorders59†††	5.1±0.7	9.2±1.0	1.9±0.2	33	8	42
	Hearing loss‡	2.3±0.5	5.0±0.7	‡‡‡‡
	Normal pregnancy and/or delivery	1.1±0.4	1.9±0.6	6.6±0.4	98	67	7
	System-level condition categories‡‡‡, §§§, ∥∥∥∥, ¶¶¶
	Diseases of the respiratory system	44.9±1.6	42.0±1.9	38.3±0.6	1	1	1
	Diseases of the musculoskeletal system and connective tissue	39.5±1.5	30.5±1.8	18.2±0.5	2	4	4
	Diseases of the digestive system	38.4±1.6	34.2±1.9	27.8±0.6	3	2	2
	Injury and poisoning	31.8±1.4	28.1±1.7	23.5±0.6	4	5	3
	Diseases of the circulatory system	28.7±1.3	24.2±1.6	9.8±0.3	5	6	9
	Diseases of the sense organs (ear and eye)∥∥∥∥	28.5±1.3	30.5±1.7	15.3±0.5	6	3	5
	Endocrine, nutritional, and metabolic diseases and immunity	22.5±1.4	14.6±1.4	8.4±0.4	7	10	12
	Diseases of the genitourinary system	21.7±1.3	21.2±1.5	13.3±0.4	8	7	7
	Other conditions¶¶¶	21.2±1.4	19.0±1.4	11.9±0.4	9	8	8
	Mental disorders	18.6±1.3	14.5±1.5	8.2±0.4	10	11	13
	Diseases of the skin and subcutaneous tissue	17.4±1.1	12.9±1.3	9.4±0.3	12	13	10
	Diseases of the nervous system∥∥∥∥	14.3±1.1	9.0±1.0	7.5±0.3	13	15	14
	Neoplasms	11.7±1.1	11.8±1.2	6.0±0.3	14	14	15
	Infectious and parasitic diseases	11.4±1.0	12.9±1.3	8.7±0.3	15	12	11

NOTE. Two additional condition categories are reported such that the top 10 incident conditions in each limitation group are shown. Incident conditions are also shown collapsed into 19 system-level condition categories.

⁎ CCS categories “Administrative/social admission,” “Medical examination/evaluation,” and the system-level category “Health services encounters” were not reported because they do not represent health conditions. † Includes conditions such as diphtheria, streptococcal sore throat, acute nasopharyngitis (common cold), etc. ‡ Most frequent disease specific ICD-9-CM code within Clinical Classification category for all limitations groups. § Includes conditions such as certain arthropathies, loose bodies in joints, pathologic/recurrent joint dislocation, joint ankylosis, and unspecified disorders of joint (eg, effusion or hemarthrosis). ∥ Includes general symptoms such as sleep disturbances, nonspecific findings on examination of blood, other nonspecific abnormal findings, and other ill-defined and unknown causes of morbidity and mortality ¶ Includes conditions such as eosinophilia myalgia syndrome, polymyalgia rheumatica, adhesive capsulitis of shoulder, rotator cuff syndrome of shoulder and allied disorders, joint enthesopathies, etc. # Includes conditions such as corns and callosities, other hypertrophic and atrophic conditions of skin, other dermatoses, diseases of nail, diseases of hair and hair follicles, disorders of sweat and sebaceous glands, etc. ⁎⁎ Includes conditions such as deviated septum, chronic pharyngitis and nasopharyngitis, allergic rhinitis, etc. †† Includes conditions such as other and unspecified neurotic disorders, certain adjustment reactions, depressive disorder (not elsewhere classified), etc. ‡‡ Includes conditions such as pulmonary congestion and hypostasis, postinflammatory pulmonary fibrosis, other alveolar and parietoalveolar pneumonopathy, pulmonary eosinophilia, etc. §§ Includes conditions such as asphyxia, late effects of cranial nerve injury or certain complications of trauma, etc. ∥∥ Includes conditions such as acute dilatation of stomach, persistent vomiting, gastroparesis, etc. ¶¶ Includes conditions such as gastroenteritis and colitis due to radiation, toxic gastroenteritis and colitis, constipation, irritable colon, functional diarrhea, peritoneal adhesions, perforation of intestine, etc. ## Includes conditions such as degenerative disorder of globe (unspecified), progressive high degenerative myopia, siderosis, other metallosis, hypotony of eye (unspecified), degenerated conditions of globe, etc. ⁎⁎⁎ Includes conditions such as phlebitis and thrombophlebitis of intracranial venous sinuses, secondary Parkinsonism, reflex sympathetic dystrophy, neuromyelitis optica, Schilder’s disease, etc. ††† Includes conditions such as disorders of external ear, unspecified disorder of middle ear and mastoid, tinnitus, disorders of acoustic nerve, otorrhea, otalgia, conductive and sensorineural hearing loss, etc. ‡‡‡ Modification of the 17 system-level code described in Elixhauser and Steiner.60 §§§ Estimates not reported due to unweighted analytic cell sizes <30 and/or relative standard errors >30%. Estimates for the categories “diseases of the blood and blood forming organs,” “complications of pregnancy, childbirth, and the puerperium,” “congenital anomalies,” and “certain conditions originating in the perinatal period” were not reported for this reason. ∥∥∥∥ Diseases of the sense organs were separated from the original category 6, diseases of the nervous system. ¶¶¶ Health services encounters were separated from the original category 17, other conditions. ### Includes symptoms such as disturbance of sensation, rash, superficial swelling, edema, unspecified jaundice, cyanosis, pallor and flushing, spontaneous ecchymosis, or induration or thickening of skin. ⁎⁎⁎⁎ Includes conditions such as achlorhydria, acute dilatation, persistent vomiting, gastroparesis, or dyspepsia. †††† Includes symptoms such as nausea and vomiting, heartburn, dysphagia, gas pain, visible peristalsis, abnormal bowel sounds, incontinence of feces, abnormal feces, or diarrhea. ‡‡‡‡ These estimates had unweighted analytic cell sizes <30 and relative standard errors >30%. They were not reported because they were considered unreliable.

When conditions were collapsed into 19 system-level categories, a similar gradient in the magnitude of incident conditions was observed (see table 2). In all reportable system-level categories, incidence rates were much higher in adults with mobility limitations compared with those without limitations. The mean number of CCS-defined incident conditions over the 2-year study period was greatest in adults with mobility limitations (mean, 4.7; 95% CI, 4.4−4.9) followed by adults with nonmobility limitations (mean, 3.9; 95% CI, 3.7−4.2) and those without limitations (mean, 2.6; 95% CI, 2.5−2.7). When added to the number of prevalent conditions existing at round 1, the mean comorbid disease burden for the entire study period in adults with mobility limitation was 8.2 (95% CI, 7.9−8.5) compared with 6.3 (95% CI, 5.9−6.6) for adults with nonmobility limitations and 3.9 (95% CI, 3.8−4.0) for adults without limitations.

In logistic regression analyses where the presence or absence of high ranking incident CCS condition categories were the outcome variables, controlling for individual characteristics and resources, significant differences in the odds of incident conditions remained across limitation groups (table 3). Compared with adults without limitations, those with mobility limitations were more likely to acquire a variety of musculoskeletal disorders, general symptoms such as sleep disturbances (residual codes), intestinal and urinary tract infections, essential hypertension, other upper and lower respiratory disease, other mental conditions, other injuries, disorders of the teeth and jaw, gastrointestinal disorders, chronic obstructive pulmonary disease (COPD) and bronchiectasis, headache (including migraine), other nervous system disorders, fluid and electrolyte disorders, and other ear and sense organ disorders. Adults with nonmobility limitations were also more likely to acquire many of the same conditions. Risk for common acute conditions such as other upper respiratory infections (including streptococcal sore throat and colds), influenza, and sprains and strains were the same across groups. Using the system-level incident condition categories as outcome variables, there were also significant differences across limitation groups based on logistic regression analyses (table 3). Compared with adults without limitations, those with mobility limitations had greater odds of incident diseases of the endocrine and metabolic system, nervous system, circulatory system, respiratory system, digestive system, genitourinary system, and musculoskeletal system, as well as mental disorders, diseases of the sense organs, diseases of the skin, injury and poisoning, and other conditions. Other factors that consistently posed an increased risk for incident conditions included female sex, living alone, and having a source of health care with difficulty obtaining care (data not shown). Results of the linear regression model predicting number of incident conditions across limitation groups (table 4) revealed that adults with mobility and nonmobility limitations were more likely to develop a greater number of incident conditions than adults without limitations, and that the number of prevalent conditions posed an additional risk for incident conditions along with other factors.

Table 3. Odds of Acquiring Incident Conditions (27/259 CCS categories59⁎ and 15 of 19 system-level categories†) Across Limitation Groups Controlling for Individual Characteristics and Resources

	Condition Categories	Odds Ratios (95% CIs)
		Mobility Limitation	Nonmobility Limitation
	CCS condition categories59⁎
	Other upper-respiratory infections	1.17(0.97−1.40)	1.09(0.89−1.35)
	Other nontraumatic joint disorders	2.31(1.81−2.97)	1.42(1.07−1.88)
	Residual codes, unclassified	1.89(1.51−2.37)	1.48(1.12−1.97)
	Intestinal infection	1.30(1.03−1.65)	1.16(0.91−1.48)
	Other connective tissue disease	1.78(1.44−2.19)	1.56(1.23−1.97)
	Spondylosis, intervertebral disk disorders, other back problems	1.82(1.42−2.32)	1.42(1.09−1.86)
	Other skin disorders	1.25(0.97−1.61)	0.99(0.73−1.34)
	Essential hypertension	1.37(1.05−1.80)	1.29(0.92−1.80)
	Other upper-respiratory disease	1.35(1.07−1.71)	1.28(0.94−1.75)
	Other mental conditions	2.01(1.53−2.64)	1.43(1.02−2.02)
	Other lower-respiratory disease	2.30(1.75−3.01)	1.19(0.82−1.74)
	Urinary tract infections	1.61(1.18−2.20)	1.48(1.07−2.04)
	Other injuries and conditions due to external causes	1.61(1.18−2.19)	1.36(0.98−1.88)
	Disorders of teeth and jaw	1.44(1.08−1.91)	0.90(0.64−1.28)
	Other disorders of stomach and duodenum	1.65(1.20−2.26)	1.71(1.24−2.36)
	COPD and bronchiectasis	1.39(1.07−1.81)	1.50(1.03−2.19)
	Other gastrointestinal disorders	1.78(1.36−2.32)	1.79(1.27−2.54)
	Headache, including migraine	1.68(1.22−2.32)	1.12(0.77−1.64)
	Sprains and strains	1.26(0.95−1.67)	1.11(0.83−1.49)
	Other eye disorders	1.38(0.95−1.98)	1.25(0.87−1.80)
	Influenza	1.28(0.99−1.65)	1.10(0.78−1.56)
	Other nervous system disorders	2.49(1.78−3.49)	1.50(0.94−2.38)
	Fluid and electrolyte disorders	1.87(1.24−2.83)	1.35(0.79−2.30)
	Other ear and sense organ disorders	1.48(1.01−2.15)	3.34(2.44−4.57)
	Normal pregnancy and/or delivery	0.63(0.29−1.38)	0.48(0.21−1.08)
	System-level condition categories†
	Infectious and parasitic diseases	1.19(0.95−1.50)	1.50(1.17−1.92)
	Neoplasms	1.24(0.98−1.57)	1.45(1.12−1.87)
	Endocrine, nutritional, and metabolic diseases and immunity	1.70(1.39−2.07)	1.26(0.98−1.63)
	Mental disorders	1.78(1.46−2.18)	1.47(1.13−1.89)
	Diseases of the nervous system	2.09(1.65−2.64)	1.29(0.97−1.71)
	Diseases of the sense organs(ear and eye)	1.22(1.03−1.44)	1.79(1.48−2.17)
	Diseases of the circulatory system	1.70(1.42−2.04)	1.88(1.54−2.29)
	Diseases of the respiratory system	1.36(1.18−1.57)	1.20(1.02−1.42)
	Diseases of the digestive system	1.50(1.29−1.75)	1.31(1.09−1.58)
	Diseases of the genitourinary system	1.48(1.23−1.78)	1.77(1.44−2.16)
	Diseases of the skin and subcutaneous tissue	1.35(1.08−1.68)	1.12(0.88−1.43)
	Diseases of the musculoskeletal system and connective tissue	1.96(1.69−2.28)	1.53(1.27−1.84)
	Injury and poisoning	1.49(1.27−1.74)	1.24(1.03−1.50)
	Other conditions	1.45(1.19−1.76)	1.44(1.17−1.78)

NOTE. “No limitation” was the reference group. Significant differences(P<.05) are in bold face.

⁎ CCS categories “Administrative/social admission,” “Medical examination/evaluation,” and the system-level category “Health services encounters” were not reported because they do not represent health conditions. † Modification of the 17 system-level code described in Elixhauser and Steiner.60 Estimates for the categories “diseases of the blood and blood forming organs,” “complications of pregnancy, childbirth, and the puerperium,” “congenital anomalies,” and “certain conditions originating in the perinatal period” were not reported because they had unweighted analytic cell sizes <30 and/or relative standard errors >30% and were considered unreliable. Diseases of the sense organs were separated from the original category 6, diseases of the nervous system. Health services encounters were separated from the original category 17, other conditions.

Table 4. Results of Linear Regression Model Predicting the Number of CCS-Defined Incident Conditions Across Limitation Groups

	Independent Variables	β ± SE	P
	Limitation group
	No limitation	Referent	Referent
	Mobility limitation	0.65±0.11	<.001
	Nonmobility limitation	0.65±0.11	<.001
	Sex
	Men	Referent	Referent
	Women	0.72±0.05	<.001
	Age (y)
	18−23	Referent	Referent
	24−29	−0.07±0.10	.47
	30−34	−0.04±0.11	.68
	35−39	−0.04±0.10	.68
	40−44	−0.10±0.10	.33
	45−49	0.03±0.11	.79
	50−54	0.02±0.12	.89
	55−59	0.21±0.14	.13
	60−64	0.15±0.15	.34
	65−69	0.59±0.17	<.001
	70−74	0.55±0.17	.002
	75−79	0.94±0.19	<.001
	80−84	0.91±0.30	.002
	≥ 85	1.07±0.43	.012
	Race and ethnicity
	Non-Hispanic white	Referent	Referent
	Non-Hispanic black	−0.69±0.07	<.001
	Hispanic	−0.19±0.08	.013
	Other	−0.35±0.12	.004
	BMI
	Not obese	Referent	Referent
	Obese	0.24±0.07	<.001
	Marital status
	Spouse in the house	Referent	Referent
	Not married/no spouse	−0.04±0.08	.60
	Family size
	Living alone	0.59±0.11	<.001
	2	0.48±0.08	<.001
	3−4	0.23±0.06	<.001
	≥5	Referent	Referent
	Education
	< High school	−0.25±0.07	.001
	High school	−0.23±0.06	<.001
	> High school	Referent	Referent
	Poverty status
	Poor (<100% poverty line)	−0.03±0.11	.77
	Low income (100%−199% poverty line)	0.03±0.08	.69
	Middle income (200%−399% poverty line)	0.04±0.06	.58
	High income (≥400% poverty line)	Referent	Referent
	Insurance coverage
	Any private insurance	Referent	Referent
	Public insurance only	−0.20±0.10	.06
	Uninsured	−0.42±0.08	<.001
	Access to health care
	Usual source/no difficulty obtaining health care	Referent	Referent
	Usual source/difficulty obtaining health care	0.75±0.10	<.001
	No usual source/no difficulty obtaining health care	−0.62±0.06	<.001
	No usual source/difficulty obtaining health care	0.29±0.17	.10
	No. of prevalent conditions	0.37±0.02	<.001

NOTE. Significant differences (P<.05) are in boldface.

Study Limitations 

To facilitate analysis, it was necessary to collapse ICD-9 codes into a manageable number of meaningful categories. For the current study, a well-recognized, peer reviewed, and published system was chosen: the Clinical Classifications for Health Policy Research developed by the AHRQ.59, 60 However, this classification was intended to provide a system for reporting hospital statistics by diagnosis. Thus, conditions frequently recorded in a hospital setting received individual codes (eg, types of cancer) whereas conditions that were unlikely to be recorded in a hospital setting (eg, a common cold) were grouped into “other” categories (eg, colds were grouped in “other upper respiratory infections”). Conditions in the “other” categories may occur frequently in the community, however. In the current study, many of the incident conditions that were reported by the community-dwelling population were grouped in these “other” categories, limiting the interpretation of the findings. This was addressed, in part, by identifying the most frequently occurring ICD-9 code that reflected a specific condition within each of the “other” CCS categories for all limitation groups. Use of the CCS for disease classification in community-based studies may not be optimal for this reason.

Self-report of health conditions imposes benefits and limitations on the study. Use of respondent reports has the advantage of adequate population coverage because persons who did not encounter the health care system or have health insurance coverage could provide information about health conditions that they had experienced. However, respondents may make reporting errors. They may not recall all of their medical conditions, may not be aware of the presence of a condition, and may not report stigmatized conditions. Respondents may not report conditions with the level of specificity of health care providers and the level of agreement between respondent and provider reports increases when conditions are collapsed into broader categories.66, 67, 68 Use of both the CCS and the less specific system-level categories for the current study was advantageous for this reason.