American Journal of Physical Medicine & Rehabilitation Articles Ahead of Print DOI: 10.1097/PHM.0000000000000848
Disability and health consequences of Traumatic Brain Injury: national prevalence
D
Claire Jourdan1,2, PhD, Philippe Azouvi1,2, PhD, François Genêt1,3,4, PhD, Nicolas Selly2, MS,
1
EP TE
Loic Josseran2,5, PhD, Alexis Schnitzler1,2, PhD
Département de Médecine Physique et de Réadaptation, Service de Médecine Physique et de
Réadaptation, CHRU Montpellier, Site Lapeyronie, 371 avenue du Doyen Gaston Giraud, 34295, Montpellier cedex 5, France 2
HANDIReSP EA 4047, Université de Versailles Saint-Quentin, UFR des Sciences de la Santé –
Simone Veil, 2 avenue de la Source de la Bièvre, 78180 Montigny-Le-Bretonneux, France CIC-IT 1429, 104, boulevard Raymond Poincaré, 92380 Garches, France
4
« End:icap » U1179 INSERM, Université Versailles Saint Quentin en Yvelines, UFR des
C
3
Sciences de la Santé – Simone Veil, 2 avenue de la Source de la Bièvre, 78180 Montigny-Le-
5
C
Bretonneux, France
Département d‟Information Médicale, Assistance Publique-Hôpitaux de Paris, Hôpital
A
Raymond Poincaré, 104, boulevard Raymond Poincaré, 92380 Garches, France
Corresponding author: Claire Jourdan. Service de Médecine Physique et de Réadaptation CHRU Montpellier, Site Lapeyronie 371 avenue du Doyen Gaston Giraud, 34295, Montpellier cedex 5, France Email: [email protected]
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Funding The Disability and Health Survey was designed, undertaken and financed by the French National Institute of Statistics (INSEE) and the Head Office of Research, Studies, Evaluation and Statistics (DREES) of the Social Affairs Ministry of France. These institutes did not finance the data analysis nor the preparation of the manuscript.
D
The data analysis was carried out with logistical help from réseau Quételet-CNRS and the Federative Institute for Research on Handicap (IFRH-25). Full access to the survey data was
EP TE
provided to the study authors.
This report received financial support from IPSEN for the data analysis and manuscript preparation. This sponsor did not participate in either the study design, the collection of the data
A
C
C
nor in the interpretation of results.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Abstract Objective To measure the prevalence of traumatic brain injury-related disability and health status in the general population. Design
D
The French National Survey, conducted in households and institutions, assessed 33,896 adults. Data included sequelae from traumatic brain injury, impairments, current health conditions and
EP TE
uses of health services. Analyses, adjusted for age and gender, compared subjects who declared sequelae from traumatic brain injury (n=479) to the remaining survey population (n=33,287). Use of weighting factors ensured that results were representative of the national population. Results
Prevalence of persistent sequelae from traumatic brain injury in France was 704/100,000. Median time since injury was 14 years. For all Core Set items of the International Classification of Functioning, subjects with traumatic brain injury reported more impairments than the control
C
population: adjusted odds ratios from 1.7 (behavioral difficulties) to 8.6 (motor difficulties). Rates of cardiovascular, respiratory, musculoskeletal, digestive, urological, neurological and
C
psychiatric conditions were higher in the traumatic brain injury population. Use of health services was greater, and women with traumatic brain injury had higher rates of unmet health
A
needs.
Conclusion
Persistent sequelae from traumatic brain injury significantly affect health in the general population. Planning of post-traumatic brain injury-care should address the chronic needs of these persons.
Keywords: “Brain injuries”; “Epidemiology”; “Prevalence”; “Disability”
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Introduction Traumatic Brain Injury (TBI) is a frequent and severe event, with an average annual incidence of 262 / 100,000 inhabitants in Europe 1. TBI causes 10.5 deaths / 100,000 inhabitants / year 1 and permanent disability for nearly half of hospitalized survivors 2. Information is, however, still lacking for the prevalence of long-term sequelae 2. Knowledge of
D
the prevalence of TBI-related disability is essential for the evaluation of its impact and the definition of health priorities 3. Studies of the Global Burden of Disease use a prevalence-based
EP TE
approach 4, as do analyses on the societal costs of brain disorders 5. Prevalences of post-TBI disability are extrapolated in these studies and data are based on incidences of hospitalizations for TBI. The most frequently cited prevalence rate of disability following TBI (2% of the population of the United States of America) was also obtained by computations 6.
However, such computations require a precise knowledge of incidence, mortality, and natural history which, for TBI, is challenging to ascertain. Variation in reported incidence rates is high (between 47 and 354 / 100,000 in European studies 1), due to variability in identification and
C
classification of cases. Most TBI incidences and outcome studies are based on hospital discharge
C
data, which do not take account of TBIs treated in the ambulatory setting. Furthermore, several groups are rarely studied: military personnel and veterans, children under 15 years of age and
A
individuals with limited access to care facilities. Moreover, while data are abundant on early TBI outcomes, the longer-term stories of persons with TBIs are poorly described in the literature, although it has been shown that impairments and dependence change over time
7,8
. Incidence-
based computations also assume epidemiological stability in terms of rate, mortality, probability of disability, and age-groups at greatest risk, which in the case of TBI are continuously evolving 9
.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Detailed information on long-term impairments, health problems and the needs of people with sequelae from TBI is necessary to plan their long-term care and management. Such information on the long-term consequences of TBI can be found within population-based, cross-sectional studies, in addition to the data from cohort studies
10
, which for TBI are particularly subject to
selection and attrition bias 11.
D
One such population-based, cross-sectional study is the French National Disability and Health Survey, a large-scale survey representative of both national households and healthcare
EP TE
institutions. The aims of the present report are (1) to assess the prevalence of persistent sequelae from TBI in the general adult population; (2) to describe impairments, health conditions and use of health care services in adults with persistent sequelae from TBI and to quantify these findings with reference to the remaining non-TBI survey sample.
Methods
C
Experimental Design
The Disability and Health Survey
12
was a national cross-sectional community-based survey
C
carried out in 2008 and 2009 by the French Statistical Institute and the French Head Office of Research, Studies, Evaluation and Statistics. The objectives of this survey were to provide
A
comprehensive data on health, disability and dependence for all causes of health loss in a representative sample of the resident population in France at that time. It evaluated two populations: people living at home and those living in healthcare institutions (see flow chart in figure 1).
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
People Living at Home For the population living at home, a two-stage sampling method was used to select a representative sample from all French territories (including overseas French departments and territories). The United Nations publication “Guidelines and Principles for the Development of Disability Statistics”13 provided methodological guidance for our data collection and analysis.
D
The first stage was a screening survey: 262,963 people living in 111,592 households were selected at random from the whole national population as assessed by the 2006 census. A short
EP TE
survey collected data on their current health status, including details of activity limitations and need for help. Respondents were categorized into four levels of disability according to this screening survey.
The second stage of data collection selected the final study sample from the initial screening survey respondents. The subjects were stratified into groups depending on their geographical location and disability level. In order to improve the representativeness of results regarding people with disabilities, disproportionate randomization was used, and higher sampling
C
coefficients were applied to those with higher levels of disability. The final study population
C
consisted of 29,931 home-living individuals. People Living in Healthcare Institutions
A
Healthcare institutions were selected at random from all French residential and medical care homes. The randomization was stratified according to seven categories of institutions and two geographical areas. The sampling probabilities were proportional to the size of the institution. A total of 1,520 institutions were selected and agreed to participate. Six to nine individuals were randomly selected among those living permanently in each institution, resulting in a sample of 9,104 institutionalized individuals.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Final Sample The present report included all adult participants from both parts of the survey although those who could not, even with help from a proxy, provide information on their medical conditions were excluded. This resulted in a final study population sample of 33,766 adults (see details on population selection on the study flow-chart of figure 1).
D
Each respondent was assigned a weighting which combined both design and non-response weights. Design weights resulted from the disproportionate sampling method (described above)
EP TE
and were based on self-reported disability and area of residence. Non-response weights depended on the probability of non-response and were estimated by logistic regression with age, sex, type of household, marital status, health and disability as independent variables. Using individual weightings, representative results could be estimated at a national level, and resulted a statistically estimated total weighted population of 49,028,826 adults living at home and in healthcare institutions across all French territory.
C
Data Collection
Data collection was financed and monitored by the French National Institute of Statistics, which
C
employed permanent, professional investigators. Investigators were provided with training on the concepts and objectives of the study and a detailed instruction manual. Face-to-face interviews
A
took place in participants‟ places of residence using computer-assisted interview formats. Interviews took on average one hour, although there was great variability depending on the health and disability of respondents. In 92.0% of cases, the respondent was the person concerned, in 3.5% of cases the respondent was aided by a proxy, and in 4.4% of cases a proxy responded on behalf of the person. The questionnaire, available on the website of the Federative Institute for Research on Disability 14, was based on the International Classification of Functioning (ICF)
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
and other World Health Organization assessments. It was structured around fourteen domains, including health status, health service use, impairments and functional limitations. In the health status section, participants were asked to identify their health problems from a list of 52 conditions; checklists were used as they have been shown to improve the quality of subject self-reporting
15
. The list (detailed in the Appendix, Supplemental Digital Content 1,
D
http://links.lww.com/PHM/A512) was preceded by the question, “Do you have, or have ever had, one of the following diseases or health problems?” The list included an item on injuries:
EP TE
“Harm or permanent sequelae caused by an accident”. Individuals who declared injury-related sequelae or harm were then asked the following question: “Are these the sequelae of a Traumatic Brain Injury”? Other items in the list provided information on other health conditions and whether they were ongoing or had occurred in the 12 months preceding the survey.
Aspects of health care and services that were clinically relevant for persons with TBI were selected from amongst the questions addressing care use in the 12 months preceding the survey. The following question addressed unmet needs for medical services: “In the past 12 months did
C
you have to give up on medical, surgical, or dental services although you really needed them?”
C
No question addressed unmet needs for rehabilitation or other non-medical services. Items from the impairment and functional sections were selected according to the Brief Core Set 16
(see the Appendix for details of selected survey questions, Supplemental
A
of the ICF for TBI
Digital Content 1, http://links.lww.com/PHM/A512). This Core Set was developed by the ICF Research Branch to define the essential domains of functioning and disability that should be considered when describing the impact of TBI on a person. It consisted of a list of categories that were selected by literature review, patient focus groups and a survey of 107 international experts and refined by an iterative decision-making process during an international consensus conference 16
. Participants were instructed to record only permanent impairments and not transitory
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
problems. A cognitive impairment was defined as a difficulty in one of the following domains: memory, understanding, concentrating, learning, temporal orientation or problem-solving. In order to describe disability levels of persons with TBI sequelae, a classification derived from the Glasgow Outcome Scale (GOS)
17
was used. The GOS is the most widely used scale for
assessing global head injury outcomes. Based on a short structured interview
18
, four levels of
D
global disability are defined: a “vegetative state” referred to a person who does not respond to instructions; “severe disability” described the inability to perform daily activities (such as
EP TE
washing, dressing or using a telephone); “moderate disability” referred to significant limitations in work, leisure or social activities, and a classification of “good recovery” indicated that there were only small disruptions to everyday life. The Disability and Health Survey did not include the structured interview for the GOS, but several questions addressed the same aspects. Using answers to 29 of the questions in the survey, a proxy GOS level of each person was determined (see
Appendix
for
survey
questions,
Supplemental
Digital
Content
1,
http://links.lww.com/PHM/A512).
C
Statistical Analyses
Statistical analyses were carried out using the SAS 9.3 statistical software (SAS Institute, Cary,
C
North Carolina). In the descriptive results, counts corresponded to the number of subjects in the survey, while frequencies, medians and quartiles were estimated with specific SAS procedures
A
for handling complex sample designs, using the strata and weighting factors described above, to provide representative results on a national level. The „Proc Surveymeans‟ and „Proc Surveyfreq‟ procedures used the Taylor expansion method to provide weighted frequencies, weighted percentages, and other descriptive analyses and variance estimates while accounting for unequal weightings.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
After assessment of the frequency of persistent sequelae from TBI in the total adult population, the population reporting TBI sequelae was described with regard to demographics, professional status (main current occupation or last occupation), context and time since injury, and disability level on the GOS scale. In approximately 3% of cases, respondents declared that sequelae resulted from several different TBIs; in these cases, the time and context of the oldest injury
D
were recorded. Most impairments, health conditions and health services considered in this report were liable to
EP TE
be frequent in the general population. Therefore, health characteristics in the sample reporting TBI sequelae (“TBI sample”) were described with reference to characteristics in the remaining survey sample (the “control” or “non-TBI sample”). Frequencies of impairments, conditions and services were assessed in both samples using „Proc Surveyfreq‟ procedures.
Statistical comparisons of frequencies in the TBI and non-TBI samples were computed after adjustment for age and gender using multiple regression models. The „Proc Surveylogistic‟ and „Proc Surveyreg‟ procedures fitted logistic and linear models and computed regression
C
coefficients and their variance while accounting for weighting factors. The dependent variables were impairment, health condition or health service and independent variables were presence of
C
sequelae from TBI (“TBI” vs “no-TBI”), age and gender. Partial Wald tests investigated the association between TBI sequelae and the health characteristic after adjustment for age and
A
gender. Given the multiplicity of analyses, the p-value limit for significance was chosen as below 0.01. To eliminate gender as a confounding variable, the interaction terms gender x TBI sequelae were systematically tested for. When this interaction term was significant, results for that data set were computed separately for men and women. Rates of missing data ranged from 0% to 0.08% for all variables except pain (40% missing), and time and context of injury (40% missing). Analyses were performed on complete cases.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Research Ethics This survey was planned as a research project and performed in collaboration with the French National Institute of Statistics. The survey was declared to be of public interest by the National Council of Statistical Information and was approved by the National Commission for Informatics and Liberty, which enforces French legislation on the collection and use of personal data (no. 78
D
– 17). According to French legislation, written consent was not required for this type of survey study; participants gave informed oral consent before performing the survey. This study
EP TE
conforms to all STROBE guidelines and reports the required information accordingly (see Checklist, Supplemental Digital Content 2, http://links.lww.com/PHM/A513). Results
Prevalence of Persistent Sequelae from TBI
In the survey sample of 33,766 adults, 479 individuals declared themselves to be living with TBI-related sequelae. The weighted prevalence of persistent TBI sequelae was 704.2 / 100,000
C
inhabitants (CI95% = [573.3 – 835.0]), corresponding to a weighted frequency of 344,495 adults living with TBI-related impairments in the total French adult population. Detailed characteristics
C
of this group are shown in table 1. The prevalence of TBI sequelae was 920.8 / 100,000 in men and 508.8 / 100,000 in women. The prevalence of TBI sequelae with advancing age (in years)
A
was 518 / 100,000 (18-40y), 882 / 100,000 (40-60y) and 725 / 100,000 (>60 y). Health, Impairments and Health Care Usage For all impairments, participants with self-reported TBI had an increased risk of having difficulties compared to those without TBI, after adjusting for age and gender (all p-values Understanding
22.3%
5.8%
4.5 [2.8-7.2]***
13.9%
19.6%
42.3%
> Concentrating
21.3%
7.8%
3.2 [2.2-4.5]***
6.5%
22.4%
46.8%
A
> Memory
(n=252)
60.5%
C
Cognitive impairments5
(n=155)
EP TE
Pain4
(n=72)
> Learning
27.4%
10.3%
3.4 [2.3-5.2]***
13.3%
29.1%
51.5%
> Temporal orientation
17.9%
5.9%
3.5 [2.3-5.4]***
4.8%
16.2%
45.1%
> Problem solving
15.7%
5.8%
3.1 [2.1-4.5]***
1.0%
10.8%
51.0%
Impulsivity or aggression
35.7%
24.6%
1.7 [1.2-2.4]*
25.8%
32.9%
59.3%
Psychological
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
impairments6 32.6%
14.8%
2.6 [1.6-4.1]***
10.5%
33.0%
73.6%
Female participants
63.3%
21.2%
6.7 [3.4-13.2]***
52.3%
65.2%
80.6%
Frequencies took unequal weightings into account in order to be representative of the French
adult population.
2
D
1
Male participants
Adjusted Odds Ratios (OR) and 95% Confidence Intervals (CI95%) were
EP TE
obtained through weighted logistic regression of the presence of the impairment, with age and gender as adjustment variables. p-values for partial Wald tests: ***:
Disability and health consequences of Traumatic Brain Injury: national prevalence
D
Claire Jourdan1,2, PhD, Philippe Azouvi1,2, PhD, François Genêt1,3,4, PhD, Nicolas Selly2, MS,
1
EP TE
Loic Josseran2,5, PhD, Alexis Schnitzler1,2, PhD
Département de Médecine Physique et de Réadaptation, Service de Médecine Physique et de
Réadaptation, CHRU Montpellier, Site Lapeyronie, 371 avenue du Doyen Gaston Giraud, 34295, Montpellier cedex 5, France 2
HANDIReSP EA 4047, Université de Versailles Saint-Quentin, UFR des Sciences de la Santé –
Simone Veil, 2 avenue de la Source de la Bièvre, 78180 Montigny-Le-Bretonneux, France CIC-IT 1429, 104, boulevard Raymond Poincaré, 92380 Garches, France
4
« End:icap » U1179 INSERM, Université Versailles Saint Quentin en Yvelines, UFR des
C
3
Sciences de la Santé – Simone Veil, 2 avenue de la Source de la Bièvre, 78180 Montigny-Le-
5
C
Bretonneux, France
Département d‟Information Médicale, Assistance Publique-Hôpitaux de Paris, Hôpital
A
Raymond Poincaré, 104, boulevard Raymond Poincaré, 92380 Garches, France
Corresponding author: Claire Jourdan. Service de Médecine Physique et de Réadaptation CHRU Montpellier, Site Lapeyronie 371 avenue du Doyen Gaston Giraud, 34295, Montpellier cedex 5, France Email: [email protected]
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Funding The Disability and Health Survey was designed, undertaken and financed by the French National Institute of Statistics (INSEE) and the Head Office of Research, Studies, Evaluation and Statistics (DREES) of the Social Affairs Ministry of France. These institutes did not finance the data analysis nor the preparation of the manuscript.
D
The data analysis was carried out with logistical help from réseau Quételet-CNRS and the Federative Institute for Research on Handicap (IFRH-25). Full access to the survey data was
EP TE
provided to the study authors.
This report received financial support from IPSEN for the data analysis and manuscript preparation. This sponsor did not participate in either the study design, the collection of the data
A
C
C
nor in the interpretation of results.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Abstract Objective To measure the prevalence of traumatic brain injury-related disability and health status in the general population. Design
D
The French National Survey, conducted in households and institutions, assessed 33,896 adults. Data included sequelae from traumatic brain injury, impairments, current health conditions and
EP TE
uses of health services. Analyses, adjusted for age and gender, compared subjects who declared sequelae from traumatic brain injury (n=479) to the remaining survey population (n=33,287). Use of weighting factors ensured that results were representative of the national population. Results
Prevalence of persistent sequelae from traumatic brain injury in France was 704/100,000. Median time since injury was 14 years. For all Core Set items of the International Classification of Functioning, subjects with traumatic brain injury reported more impairments than the control
C
population: adjusted odds ratios from 1.7 (behavioral difficulties) to 8.6 (motor difficulties). Rates of cardiovascular, respiratory, musculoskeletal, digestive, urological, neurological and
C
psychiatric conditions were higher in the traumatic brain injury population. Use of health services was greater, and women with traumatic brain injury had higher rates of unmet health
A
needs.
Conclusion
Persistent sequelae from traumatic brain injury significantly affect health in the general population. Planning of post-traumatic brain injury-care should address the chronic needs of these persons.
Keywords: “Brain injuries”; “Epidemiology”; “Prevalence”; “Disability”
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Introduction Traumatic Brain Injury (TBI) is a frequent and severe event, with an average annual incidence of 262 / 100,000 inhabitants in Europe 1. TBI causes 10.5 deaths / 100,000 inhabitants / year 1 and permanent disability for nearly half of hospitalized survivors 2. Information is, however, still lacking for the prevalence of long-term sequelae 2. Knowledge of
D
the prevalence of TBI-related disability is essential for the evaluation of its impact and the definition of health priorities 3. Studies of the Global Burden of Disease use a prevalence-based
EP TE
approach 4, as do analyses on the societal costs of brain disorders 5. Prevalences of post-TBI disability are extrapolated in these studies and data are based on incidences of hospitalizations for TBI. The most frequently cited prevalence rate of disability following TBI (2% of the population of the United States of America) was also obtained by computations 6.
However, such computations require a precise knowledge of incidence, mortality, and natural history which, for TBI, is challenging to ascertain. Variation in reported incidence rates is high (between 47 and 354 / 100,000 in European studies 1), due to variability in identification and
C
classification of cases. Most TBI incidences and outcome studies are based on hospital discharge
C
data, which do not take account of TBIs treated in the ambulatory setting. Furthermore, several groups are rarely studied: military personnel and veterans, children under 15 years of age and
A
individuals with limited access to care facilities. Moreover, while data are abundant on early TBI outcomes, the longer-term stories of persons with TBIs are poorly described in the literature, although it has been shown that impairments and dependence change over time
7,8
. Incidence-
based computations also assume epidemiological stability in terms of rate, mortality, probability of disability, and age-groups at greatest risk, which in the case of TBI are continuously evolving 9
.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Detailed information on long-term impairments, health problems and the needs of people with sequelae from TBI is necessary to plan their long-term care and management. Such information on the long-term consequences of TBI can be found within population-based, cross-sectional studies, in addition to the data from cohort studies
10
, which for TBI are particularly subject to
selection and attrition bias 11.
D
One such population-based, cross-sectional study is the French National Disability and Health Survey, a large-scale survey representative of both national households and healthcare
EP TE
institutions. The aims of the present report are (1) to assess the prevalence of persistent sequelae from TBI in the general adult population; (2) to describe impairments, health conditions and use of health care services in adults with persistent sequelae from TBI and to quantify these findings with reference to the remaining non-TBI survey sample.
Methods
C
Experimental Design
The Disability and Health Survey
12
was a national cross-sectional community-based survey
C
carried out in 2008 and 2009 by the French Statistical Institute and the French Head Office of Research, Studies, Evaluation and Statistics. The objectives of this survey were to provide
A
comprehensive data on health, disability and dependence for all causes of health loss in a representative sample of the resident population in France at that time. It evaluated two populations: people living at home and those living in healthcare institutions (see flow chart in figure 1).
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
People Living at Home For the population living at home, a two-stage sampling method was used to select a representative sample from all French territories (including overseas French departments and territories). The United Nations publication “Guidelines and Principles for the Development of Disability Statistics”13 provided methodological guidance for our data collection and analysis.
D
The first stage was a screening survey: 262,963 people living in 111,592 households were selected at random from the whole national population as assessed by the 2006 census. A short
EP TE
survey collected data on their current health status, including details of activity limitations and need for help. Respondents were categorized into four levels of disability according to this screening survey.
The second stage of data collection selected the final study sample from the initial screening survey respondents. The subjects were stratified into groups depending on their geographical location and disability level. In order to improve the representativeness of results regarding people with disabilities, disproportionate randomization was used, and higher sampling
C
coefficients were applied to those with higher levels of disability. The final study population
C
consisted of 29,931 home-living individuals. People Living in Healthcare Institutions
A
Healthcare institutions were selected at random from all French residential and medical care homes. The randomization was stratified according to seven categories of institutions and two geographical areas. The sampling probabilities were proportional to the size of the institution. A total of 1,520 institutions were selected and agreed to participate. Six to nine individuals were randomly selected among those living permanently in each institution, resulting in a sample of 9,104 institutionalized individuals.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Final Sample The present report included all adult participants from both parts of the survey although those who could not, even with help from a proxy, provide information on their medical conditions were excluded. This resulted in a final study population sample of 33,766 adults (see details on population selection on the study flow-chart of figure 1).
D
Each respondent was assigned a weighting which combined both design and non-response weights. Design weights resulted from the disproportionate sampling method (described above)
EP TE
and were based on self-reported disability and area of residence. Non-response weights depended on the probability of non-response and were estimated by logistic regression with age, sex, type of household, marital status, health and disability as independent variables. Using individual weightings, representative results could be estimated at a national level, and resulted a statistically estimated total weighted population of 49,028,826 adults living at home and in healthcare institutions across all French territory.
C
Data Collection
Data collection was financed and monitored by the French National Institute of Statistics, which
C
employed permanent, professional investigators. Investigators were provided with training on the concepts and objectives of the study and a detailed instruction manual. Face-to-face interviews
A
took place in participants‟ places of residence using computer-assisted interview formats. Interviews took on average one hour, although there was great variability depending on the health and disability of respondents. In 92.0% of cases, the respondent was the person concerned, in 3.5% of cases the respondent was aided by a proxy, and in 4.4% of cases a proxy responded on behalf of the person. The questionnaire, available on the website of the Federative Institute for Research on Disability 14, was based on the International Classification of Functioning (ICF)
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
and other World Health Organization assessments. It was structured around fourteen domains, including health status, health service use, impairments and functional limitations. In the health status section, participants were asked to identify their health problems from a list of 52 conditions; checklists were used as they have been shown to improve the quality of subject self-reporting
15
. The list (detailed in the Appendix, Supplemental Digital Content 1,
D
http://links.lww.com/PHM/A512) was preceded by the question, “Do you have, or have ever had, one of the following diseases or health problems?” The list included an item on injuries:
EP TE
“Harm or permanent sequelae caused by an accident”. Individuals who declared injury-related sequelae or harm were then asked the following question: “Are these the sequelae of a Traumatic Brain Injury”? Other items in the list provided information on other health conditions and whether they were ongoing or had occurred in the 12 months preceding the survey.
Aspects of health care and services that were clinically relevant for persons with TBI were selected from amongst the questions addressing care use in the 12 months preceding the survey. The following question addressed unmet needs for medical services: “In the past 12 months did
C
you have to give up on medical, surgical, or dental services although you really needed them?”
C
No question addressed unmet needs for rehabilitation or other non-medical services. Items from the impairment and functional sections were selected according to the Brief Core Set 16
(see the Appendix for details of selected survey questions, Supplemental
A
of the ICF for TBI
Digital Content 1, http://links.lww.com/PHM/A512). This Core Set was developed by the ICF Research Branch to define the essential domains of functioning and disability that should be considered when describing the impact of TBI on a person. It consisted of a list of categories that were selected by literature review, patient focus groups and a survey of 107 international experts and refined by an iterative decision-making process during an international consensus conference 16
. Participants were instructed to record only permanent impairments and not transitory
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
problems. A cognitive impairment was defined as a difficulty in one of the following domains: memory, understanding, concentrating, learning, temporal orientation or problem-solving. In order to describe disability levels of persons with TBI sequelae, a classification derived from the Glasgow Outcome Scale (GOS)
17
was used. The GOS is the most widely used scale for
assessing global head injury outcomes. Based on a short structured interview
18
, four levels of
D
global disability are defined: a “vegetative state” referred to a person who does not respond to instructions; “severe disability” described the inability to perform daily activities (such as
EP TE
washing, dressing or using a telephone); “moderate disability” referred to significant limitations in work, leisure or social activities, and a classification of “good recovery” indicated that there were only small disruptions to everyday life. The Disability and Health Survey did not include the structured interview for the GOS, but several questions addressed the same aspects. Using answers to 29 of the questions in the survey, a proxy GOS level of each person was determined (see
Appendix
for
survey
questions,
Supplemental
Digital
Content
1,
http://links.lww.com/PHM/A512).
C
Statistical Analyses
Statistical analyses were carried out using the SAS 9.3 statistical software (SAS Institute, Cary,
C
North Carolina). In the descriptive results, counts corresponded to the number of subjects in the survey, while frequencies, medians and quartiles were estimated with specific SAS procedures
A
for handling complex sample designs, using the strata and weighting factors described above, to provide representative results on a national level. The „Proc Surveymeans‟ and „Proc Surveyfreq‟ procedures used the Taylor expansion method to provide weighted frequencies, weighted percentages, and other descriptive analyses and variance estimates while accounting for unequal weightings.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
After assessment of the frequency of persistent sequelae from TBI in the total adult population, the population reporting TBI sequelae was described with regard to demographics, professional status (main current occupation or last occupation), context and time since injury, and disability level on the GOS scale. In approximately 3% of cases, respondents declared that sequelae resulted from several different TBIs; in these cases, the time and context of the oldest injury
D
were recorded. Most impairments, health conditions and health services considered in this report were liable to
EP TE
be frequent in the general population. Therefore, health characteristics in the sample reporting TBI sequelae (“TBI sample”) were described with reference to characteristics in the remaining survey sample (the “control” or “non-TBI sample”). Frequencies of impairments, conditions and services were assessed in both samples using „Proc Surveyfreq‟ procedures.
Statistical comparisons of frequencies in the TBI and non-TBI samples were computed after adjustment for age and gender using multiple regression models. The „Proc Surveylogistic‟ and „Proc Surveyreg‟ procedures fitted logistic and linear models and computed regression
C
coefficients and their variance while accounting for weighting factors. The dependent variables were impairment, health condition or health service and independent variables were presence of
C
sequelae from TBI (“TBI” vs “no-TBI”), age and gender. Partial Wald tests investigated the association between TBI sequelae and the health characteristic after adjustment for age and
A
gender. Given the multiplicity of analyses, the p-value limit for significance was chosen as below 0.01. To eliminate gender as a confounding variable, the interaction terms gender x TBI sequelae were systematically tested for. When this interaction term was significant, results for that data set were computed separately for men and women. Rates of missing data ranged from 0% to 0.08% for all variables except pain (40% missing), and time and context of injury (40% missing). Analyses were performed on complete cases.
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Research Ethics This survey was planned as a research project and performed in collaboration with the French National Institute of Statistics. The survey was declared to be of public interest by the National Council of Statistical Information and was approved by the National Commission for Informatics and Liberty, which enforces French legislation on the collection and use of personal data (no. 78
D
– 17). According to French legislation, written consent was not required for this type of survey study; participants gave informed oral consent before performing the survey. This study
EP TE
conforms to all STROBE guidelines and reports the required information accordingly (see Checklist, Supplemental Digital Content 2, http://links.lww.com/PHM/A513). Results
Prevalence of Persistent Sequelae from TBI
In the survey sample of 33,766 adults, 479 individuals declared themselves to be living with TBI-related sequelae. The weighted prevalence of persistent TBI sequelae was 704.2 / 100,000
C
inhabitants (CI95% = [573.3 – 835.0]), corresponding to a weighted frequency of 344,495 adults living with TBI-related impairments in the total French adult population. Detailed characteristics
C
of this group are shown in table 1. The prevalence of TBI sequelae was 920.8 / 100,000 in men and 508.8 / 100,000 in women. The prevalence of TBI sequelae with advancing age (in years)
A
was 518 / 100,000 (18-40y), 882 / 100,000 (40-60y) and 725 / 100,000 (>60 y). Health, Impairments and Health Care Usage For all impairments, participants with self-reported TBI had an increased risk of having difficulties compared to those without TBI, after adjusting for age and gender (all p-values Understanding
22.3%
5.8%
4.5 [2.8-7.2]***
13.9%
19.6%
42.3%
> Concentrating
21.3%
7.8%
3.2 [2.2-4.5]***
6.5%
22.4%
46.8%
A
> Memory
(n=252)
60.5%
C
Cognitive impairments5
(n=155)
EP TE
Pain4
(n=72)
> Learning
27.4%
10.3%
3.4 [2.3-5.2]***
13.3%
29.1%
51.5%
> Temporal orientation
17.9%
5.9%
3.5 [2.3-5.4]***
4.8%
16.2%
45.1%
> Problem solving
15.7%
5.8%
3.1 [2.1-4.5]***
1.0%
10.8%
51.0%
Impulsivity or aggression
35.7%
24.6%
1.7 [1.2-2.4]*
25.8%
32.9%
59.3%
Psychological
Copyright © 2017 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
impairments6 32.6%
14.8%
2.6 [1.6-4.1]***
10.5%
33.0%
73.6%
Female participants
63.3%
21.2%
6.7 [3.4-13.2]***
52.3%
65.2%
80.6%
Frequencies took unequal weightings into account in order to be representative of the French
adult population.
2
D
1
Male participants
Adjusted Odds Ratios (OR) and 95% Confidence Intervals (CI95%) were
EP TE
obtained through weighted logistic regression of the presence of the impairment, with age and gender as adjustment variables. p-values for partial Wald tests: ***:
