Injurious Falls Among Middle Aged and Older Adults With Multiple Sclerosis

Selection and Description of Participants 

The original project was reviewed and approved by the institutional review board of the University of Illinois at Chicago, Chicago, IL. All procedures were approved for use, including contact and recruitment materials.

Participants were recruited through the North American Research Committee on Multiple Sclerosis (NARCOMS) Registry, a volunteer registry of more than 25,000 people with MS.31 Seven hundred people with MS who were aged 55 years or older and living in the United States were randomly selected and invited by mail to participate in the study. The decision to invite the participation of 700 people was based on the original aims of the study (ie, to inform development of a fall prevention program for people with MS by collecting information regarding fall risk factors) and on previous experience with recruitment through the NARCOMS registry. Interested people were asked to contact the study office to confirm eligibility (ie, ≥55y of age, self-reported diagnosis of MS, willingness to participate in a telephone interview). The study was open to all people who met these criteria; no restrictions were placed based on location of residence (eg, community vs institution).

Of the 700 letters sent, 5 were returned because of change of address or death. A total of 384 people (55%) expressed initial interest in participating in the study, but across these people, 9 declined to participate on contact, and 6 were ineligible on screening. Across the remaining 369 people, 354 completed a full interview independently (participation rate, 51%). Fifteen people were unable to complete the full interview because of fatigue; time constraints; or cognitive impairment, which was evaluated by using the Blessed Orientation-Concentration-Memory (BOMC) test.32 These same reasons may also explain why other people with MS did not respond to the study invitation. Alternatively, it is possible that discussions about falls could raise concerns about fall risk or lead to recollection of unpleasant memories about fall-related experiences. Thus, some people may have turned down the invitation to participate in the study to avoid such conversations. A basic descriptive profile of the 354 participants who were included in the analysis of this study is provided in table 1.

Table 1. Basic Demographic Characteristics (N=354)

	Characteristic	n	%
	Age (y)⁎	66.8	7.1
	Sex
	Male	118	33.3
	Female	236	66.7
	MS-related characteristics
	Years since diagnosis⁎	21.6	12.5
	MS status in past year
	Stable or improving	125	35.4
	Deteriorating	142	40.2
	Variable	86	24.4
	MS course over the past year
	Unpredictable	44	12.8
	Mild and intermittent	102	29.6
	Steady and progressive	162	47.0
	Initially mild and intermittent then steady and progressive	37	10.7
	No. of MS-related symptoms⁎	6.1	1.9
	Health-related characteristics
	BOMC test score⁎†	2.2	2.0
	No. of health conditions (excluding MS and osteoporosis)⁎	1.6	1.3
	Housing accommodations
	Type of housing
	Independent living	324	91.5
	Nursing home or assisted living facility	8	2.3
	Other	22	6.2
	Accessibility of the home
	Bad or poor	16	4.5
	Fair	47	13.3
	Good	100	28.2
	Very good	191	54.0

⁎ Indicates a continuous variable and mean and standard deviation (SD) were presented instead of frequency and percentage. † Score range from 0 (no cognitive impairment) to 12 (severe cognitive impairment).

Data Collection Procedures 

Participants were contacted by telephone and interviewed by the principal investigator, by the co-principal investigator, or by 1 of 4 trained research assistants. Interviews lasted an average of 39 minutes (median, 35min). A structured interview guide was used to gather sociodemographic information, background about the participant's MS course, presence and extent of MS symptoms, presence of activity limitations, the use of assistive technology to aid in day-to-day tasks, availability of social support, experience of fear of falling, and falls self-efficacy.

Most of the 58 questions on the interview guide were closed-ended survey items. Wherever possible, these items were drawn from existing studies of falls or MS. In addition, 2 open-ended questions were asked. The first inquired about the participant's most recent fall. In the second, participants were asked whether they had ever sought any type of medical attention because of a fall-related injury. Definitions for injurious and noninjurious falls were not provided to participants as part of the interview. Participants who reported that they had received fall-related medical attention were asked follow-up questions about the nature of the injury and the care received. Information from these questions was recorded as fieldnotes in preparation for content analysis.

Data Analysis 

Survey data were entered into SPSS DataBuildera software and then imported into SASb software for analysis. These data were analyzed descriptively to identify potentially invalid responses and possible data entry errors. Errors were checked and corrected against the raw data on the hard-copy interview guides.

The outcome variable for the analysis was constructed by using the participant's responses to the following 2 questions: “Have you ever needed to seek any type of medical attention or care because of a fall-related injury?” (yes, no) and (if yes) “When was the most recent time you received care for a fall-related injury?” Responses were used to create a 3-level outcome variable: never experienced an injurious fall, received medical attention for an injurious fall more than 6 months ago, and received medical attention for an injurious fall within the past 6 months. These categories were selected to ensure cell size for analysis purposes. Given the ordinal nature of this outcome, proportional odds models were used to identify potential factors that may be associated with increased odds of experiencing a recent injurious fall.33

Based on knowledge about falls among people with MS14, 15 and risk factors for injurious falls among community-dwelling older adults, 12 covariates were initially considered for the analysis. These variables included: age, sex, MS status (stable or improving, deteriorating, variable), poor concentration or forgetfulness (problem, not a problem), balance or mobility (problem, not a problem), bladder control (eg, urgency, incontinence; problem, not a problem), osteoporosis (yes, no), cane and/or walker use (use, do not use), wheelchair use (use, do not use), report of taking 3 or more medications (yes, no), fall frequency (less than twice a year, each 5 to 7 weeks to twice a year, once a month or more), and fear of falling (yes, no). The a priori hypothesis was that older age, presence of osteoporosis, problems with balance and mobility, problems with concentration or forgetfulness, and greater fall frequency would be associated with increased odds of receiving medical care for an injurious fall. We were uncertain whether any other variables would be related to injurious falls and therefore assumed no relationship.

Model selection was performed on this set of 12 variables by using 3 different model selection methods: forward, backward, and a combination method using both forward and backward (ie, stepwise). For all models, a P value criterion of .05 was used for both entry into the model and remaining in it. The models obtained from the 3 methods were compared and found to be identical.

The use of a proportional odds model requires that the proportional odds assumption be satisfied. A score test for the proportional odds assumption on the final model resulted in a chi-square of χ22 equal to 2.64 and a P value of .267, thus satisfying the required assumption. The deviance (P=.533) and Pearson (P=.531) goodness-of-fit tests indicated that the final model fitted the data well.

Qualitative data from the open-ended questions were recorded as fieldnotes and were analyzed by using standard content analysis techniques.34 Four members of the research team independently analyzed the data and identified classifications or groupings of injury types. The analyses were compared, and when differences were found, discussions were held among those four until a consensus was reached. Frequencies were used to summarize the nature and number of injuries that were reported in each of the classifications developed through this process.

Description of the Sample 

A general profile of the 354 participants is provided in table 1. Additional descriptive data about the participants (data not shown in table 1) indicate that their most common symptoms were problems with balance and mobility (93.7% [n=329]), leg weakness (90.9% [n=321]), fatigue (88.4% [n=311]), problems with bladder control (78.1% [n=274]), poor concentration or forgetfulness (65.2% [n=229]), problems with bowel irregularity (59.3% [n=210]), pain (56.5% [n=199]), depression (43.1% [n=152]), and problems with seeing (38.9% [n=136]). Other than MS, participants reported living with a variety of other health conditions that influenced their health and ability to manage day-to-day activities. Some of these conditions included arthritis (37.4% [n=130]), diabetes (35.7% [n=126]), high or low blood pressure (32.9% [n=115]), visual problems (27.1% [n=95]), and heart problems (19% [n=67]).

Sixty-four percent (n=225) of the study participants reported that they experience at least 2 falls each year; 30% (n=107) reported experiencing a fall once a month or more. Among study participants who reported receiving medical care for a fall-related injury (n=177 [50%]), 41 (23%) of these people indicated that they done so in the 6 months before the interview. Descriptive details of the participants according to time of the most recent fall-related injury (ie, never, ≤6mo ago, >6mo ago) are provided in table 2. The types of injuries reported (based on open-ended questions) are shown in table 3. Fractures were the most commonly reported injury. Only 24 participants (14%) reported sustaining greater than 1 injury during their most recent injurious fall. Among those people, 1 reported greater than 1 fracture, and 2 people reported greater than 1 injury categorized as miscellaneous.

Table 2. Variables Considered in the Proportional Odds Regression Modeling for the Most Recent Fall That Required Medical Attention (N=354)

	Characteristic	Most Recent Fall That Required Medical Attention
		Never (n=169 [48.8%])		>6 Months (n=136 [39.3%])		≤6 Months (n=41 [11.9%])
		n	%	n	%	n	%
	Age (y)⁎	67.3	7.1	66.2	7.3	67.2	6.1
	Sex
	Male	70	41.4	37	27.2	11	26.8
	Female	99	58.6	99	72.8	30	73.2
	MS-related characteristics
	MS status in past year
	Stable or improving	66	39.3	43	31.6	11	26.8
	Deteriorating	63	37.5	55	40.4	21	51.2
	Variable	39	23.2	38	27.9	9	22.0
	MS symptom: poor concentration or forgetfulness
	Is a problem	107	64.1	93	68.4	26	65.0
	Not a problem	60	35.9	43	31.6	14	35.0
	MS symptom: balance or mobility problems
	Is a problem	156	93.4	125	92.6	40	97.6
	Not a problem	11	6.6	10	7.4	1	2.4
	MS symptom: bladder control problems
	Is a problem	130	76.9	105	78.9	33	80.5
	Not a problem	39	23.1	28	21.1	8	19.5
	Health-related characteristics
	Osteoporosis status
	Yes	33	20.1	47	35.1	12	30.0
	No	131	79.9	87	64.9	28	70.0
	Mobility aid: cane or walker
	Use mobility aid	136	80.5	99	72.8	29	70.7
	Do not use mobility aid	33	19.5	37	27.2	12	29.3
	Mobility aid: manual or power wheelchair
	Use mobility aid	131	77.5	102	75.0	27	65.9
	Do not use mobility aid	38	22.5	34	25.0	14	34.1
	Amount of prescription medication taken
	>3	126	75.4	108	80.0	35	85.4
	<3	41	24.6	27	20.0	6	14.6
	Fall-related characteristics
	Fall frequency
	Less than twice a year	61	36.5	51	37.8	9	22.0
	More than twice a year, but less than once a month	59	35.3	41	30.4	16	39.0
	Once a month or more	47	28.1	43	31.9	16	39.0
	Fear of falling
	Yes	80	47.6	83	61.0	29	70.7
	No	88	52.4	53	39.0	12	29.3

NOTE. Eight participants did not provide appropriate information about their most recent fall and were thus excluded from this table and subsequent analysis.

⁎ Indicates a continuous variable and mean and SD were presented instead of frequency and percentage.

Table 3. Type and Frequency of Fall-Related Injuries Among Adults (n=177) With MS, Aged 55 Years and Older

	Type of Injury Reported	Frequency
	Fracture	80
	Hip	(12)
	Lower extremity (other than hip)	(40)
	Upper extremity	(17)
	Other	(11)
	Soft tissue injuries (eg, bruises, lacerations/abrasions not requiring stitches, strains, sprains, dislocations)	60
	Laceration requiring stitches or staples	21
	Head injury (eg, subdural hematoma, concussion, “bumped” head)	9
	Other	24

NOTE. Of 177 participants with injuries, 24 reported more than 1 injury. Values in parentheses are subtotals.

Factors Associated With Increased Likelihood of Experiencing a Recent Fall 

Table 4 shows the parameter estimate, P values, and odds ratios (ORs) from the final proportional odds model obtained by examining factors associated with receiving medical attention for an injurious fall. The results of the analysis found only 2 factors, fear of falling (OR=1.94; 95% confidence interval [CI], 1.27–2.96) and osteoporosis (OR=1.65; 95% CI, 1.03–2.62), to be associated with increased likelihood of seeking medical attention for a fall-related injury within the past 6 months. On the basis of the findings, the hypothesis that older age, osteoporosis, problems with balance and mobility, problems with concentration or forgetfulness, and greater fall frequency would be associated with an increased likelihood of receiving care for a fall-related injury was partially accepted. Within this set of factors, only osteoporosis was significant in the final model.

Table 4. Results From Proportional Odds Regression Modeling the Most Recent Fall That Required Medical Attention

	Effect	df	Estimate	SE	χ2	OR	95% CI for OR	P
	Intercept (reference: never had an injurious fall)	2
	Had an injurious fall >6mo ago	1	−0.437	.171	6.52	NA	NA	.011
	Had an injurious fall in past 6mo	1	−2.579	.230	125.73	NA	NA	<.001
	Fear of falling (reference: no)	1	0.662	.216	9.37	1.94	1.27–2.96	.002
	Osteoporosis status (reference: no)	1	0.498	.237	4.42	1.65	1.03–2.62	.036

Abbreviations: NA, not applicable; SE, standard error.

Study Limitations 

Although this study contributes to and enhances the literature on fall risk among people with MS, it is limited by both its cross-sectional design and its self-report format. Prospective studies involving people with MS are needed to track actual falls, related injuries, and fall risk factors over time. Validating injuries against medical records or coding injuries by using measures of severity, such as the Injury Severity Score or the Modified Anatomic Profile that are based on the International Classification of Diseases, Ninth Revision,52 in future studies is also recommended. Using self-reported information about the receipt of medical care for a fall-related injury may have influenced the study's results, either through under- or overreporting or through inaccurate descriptions of the sustained injuries. In addition to validating injuries, future studies may also benefit from adding performance-based measures of balance and the use of explicit theoretical models of health service utilization to guide the research design and analytic process, for example, the Behavioral Model of Health Service Utilization.53, 54, 55, 56 Using such models would enable future researchers to incorporate factors that may predispose and enable the use of health services, such as proximity to medical care, availability of health insurance, and presence of a caregiver.

Because sample selection was drawn from a voluntary registry, we cannot be assured that our sample is representative of all adults aged 55 years and older with MS. Patients who respond to recruitment efforts associated with the NARCOMS registry may be somehow different than those who do not. In addition, it is possible that the people volunteering for this study are more or less likely to fall than is the general MS population of people aged 55 years and older. However, the demographics of the MS patients participating in the NARCOMS patient registry closely match those seen in the population-based study Portrait of MS that was completed by the National Multiple Sclerosis Society in 1993,20 the New York State Registry data,57 and the National Health Interview Survey data for MS.58 Our sample was also comparable to samples described of people with MS who were aged 65 years or older, with respect to average age, sex distribution, most common symptoms experienced, duration of MS, and living arrangements.59, 60, 61

The finding that 83% of our sample described the accessibility of their home environment as good or very good suggests that study participants may have been proactive in addressing home hazards. Alternatively, participants may have been unable to properly evaluate fall risks in the home. Altered internal standards or values associated with home accessibility may account for the high ratings. Such response shifts have been widely documented in literature describing measurement of quality of life among disabled people.62 Overall, our ability to assess environmental risk factors associated with fall-related injury was limited. Future studies exploring the relationship between an individual's ability to negotiate hazards in the home and his or her risk for fall-related injury are needed.