What Factors Influence The Assistive Tecnologies Provision Service In A Developing Country: The Bogota D.C. Case

Adriana M. Ríos Rincón, Antonio Miguel Cruz, Mayra R. Guarín, Pedro Sebastián Caycedo Villarraga, and María Cristina Vargas

School of Medicine and Health Sciences, Universidad del Rosario, Bogotá D.C.

ABSTRACT

The goal of this study was to determine if the provision of specific types of assistive technology (AT) devices depends on the clients’ demographic and clinical characteristics.  A retrospective cross-sectional study design was conducted over a database of AT provided to people with disability in Bogota in the period 2000-2005. The results of this study showed that the type of AT provided depended on the diagnostic type, the type of impairment, the client´s age, whether the client has care giver, client’s socio economic strata, and thegeographic zone in which the client lives because these factors. Also, the diagnostic type and the type of impairment had the strongest association level with the type of AT device provided. This study may help to all stakeholders who provide AT in taking decisions regardingdeveloping public policy and clinical actions (e.g. client’s evaluation process) for the AT provision process in Bogota.

BACKGROUND

 The Assistive Technology (AT) provision process has been recognized as a demanding issue for the rehabilitation system (Bernd, Van Der Pijl, & De Witte, 2009). Studying the AT provision process is recognized as the most critical aspect for cost-containment and health care policy (Hubbard, et al., 2010). However, studies in AT has been commonly focused on the technical aspects of the AT devices and accessibility (Bernd, Van Der Pijl, & De Witte, 2009), but studies related to the AT provision process are scarce (Bernd, Van Der Pijl, & De Witte, 2009; Hubbard-Winkler, et al., 2010). For example, two of the few studies in AT provision found that the factors associated with the provision of wheeled mobility equipment were: age, sex, race/ethnicity, diagnosis and number of comorbidities(Hubbard, Fitzgerald, Reker, Boninger, Cooper, & Kazis, 2006); and diagnosis, impairment, gender and age, and living setting (Karmarkar, Diciano, Graham, Cooper, A, & Cooper, 2012). In another study, Hubbard and colleagues (2010) studied the variation in provision of many types of AT devices in veterans poststroke. They found that administrative regions had a strong association with the AT provision, while diagnosis, demographics and severity were poorly associated with the AT provision

Even though the AT provision process has been little studied in developed countries, much less attention has been given to the factors associated to the AT provision process in low- and middle-income countries. Therefore, studying this problem is critical because these countries face specific problems such as lack of resources for disabled people (Jefferds, et al., 2010), and social issues such as poverty and violence.

Bogota, the Colombian capital, started the AT provision process in 2000 through the service AT Banks (BAT from its acronym in Spanish) which are managed by the public hospitals throughout the city. Therefore, AT provision process is carried out by the public hospitals. In order to unify and organize the scheme of AT provision process in the city, a guideline for AT provision was developed in 2007. As a result, a database that unified the information from each BAT was created (see Data and Study Sample section for more details). However, so far the factors associated with the AT devices provision to people with disability are unknown.  Therefore, the logical next step was to identify patterns in the provision process of BAT. By doing so, this research had some practical implications. For example, knowing the patterns in the provision process allows policy makers to address public policy regarding the AT provision. Also this research contributed to identify actions that can be improved, in order to ensure the equal access to the AT devices by people with disabilities. To the best of our knowledge this is the first study that examine the factors associated to the AT provision low income country.

PURPOSE

The purpose of this study was to answer the following research questions:

  • ¿Does provision of specific types of AT devices depends on the clients’ demographic and clinical characteristics?
  • Are the clients’ demographic and clinical characteristics associated with provision of specific types of AT devices?

METHOD

Study Design, Data and Study Sample

We used a retrospective cross-sectional study design in our research. We used the BAT’s clients of Bogota 2000-2005 Database. This database is housed at the District Health Secretary (SDS for its acronym in Spanish) and was developed in 2007 by Universidad del Rosario from the clinical records of the 11 public hospitals that had BATs services (covering 15/20 localities across Bogota). The database had a total of 1182 observations and 27 fields which includes client’s demographic and clinical information, the type and number of AT provided, and hospital’s administrative information. For this project we used nine fields from the original database, additionally we generated two new fields from the original database (social economic strata and geographic zone). Data are described in the next section.

Operational Definitions

Outcome (dependent) variables

In our study we used the AT device type provided (CodeATVD5C) as outcome variable. The AT device was categorized according to the ISO 9999:2006 classification (ISO/FDIS, 2006). The resulting categories were: (0) Wheelchairs (only one type of Wheelchairs was provided, i.e. standard manual wheelchairs); (1) AT for walking (including canes, walkers, and crutches); (2) Orthoses and prostheses (including: external orthoses and prostheses for upper and lower limbs); (3) AT for hearing (only one type of AT for hearing was provided, i.e. In the Ear Hearing Aid); and (4) Other ATs (including seat cushions and underlays, tactile sticks, restraint systems for wheelchairs,  bath/shower chairs, and manual Braille writing equipment).

Independent variables

We included as measures of clients’ clinical characteristics: the diagnostic type (DiagCode7C) according to the International diseases classification (ICD-10) (WHO, 2010), and the client’s type of impairment (ImparCode3C) according to the international classification of impairments, disabilities and handicaps (ICIDH) (WHO, 1980)1. As measures of clients’ demographic characteristics we included: the client’s age (AgeCode4C), and the gender (GenderCode). Additionally, we included whether the client has care giver (Caregiver), the client’s socio economic strata (EcoStraCode3C) according to the Colombian Socioeconomic stratification Act 142 1994 (DANE, 1994), thegeographic zone where the client lives (ZNCode3C), the client`s healthcare affiliation type (AfilTypeCode2C),the year of AT provision (YearCode3C), and the total number of AT devices (NprovAT) provided to each the client.

Dichotomous were coded as “0” or “1”. Each polychromous variable was represented by a set of pair’s number/category. All variables were treated as categorical. See Table 1 for more details

Statistical analysis

The data were examined in the following steps: (1) a basic univariate analysis (i.e. frequency analysis); (2) analysis of missing data and imputation procedures; (3) steep one was repeated again; (4) next, to respond to the research question 1, we performed bivariate analysis consisting in the Chi-squared independence test; and to respond to the research question 2, we obtained the Pawlik’s Contingency Coefficient. We also obtained the Cramér’s V or Phi coefficient depending on whether the independent variables were dichotomous or polychromous. A level of p<0.05 was used to identify statistically significant independence test. The statistical analysis (including data imputation procedures) was conducted in IMB SPSS software, version 20.0.

Missing data analysis and data imputation procedure

We used the missing data analysis at random assumptions approach (MAR). Therefore, we used multivariate imputation method based on logistic regression algorithms. This method is the most commonly used, flexible, and useful when multiple variables have missing values and when variables are categorical (Barnard & Meng, 1999). We used the imputation method when the percentage of missing data of a certain variable was less than or equal to 5%, otherwise the variable with all their observations were eliminated from database (Finch, 2010). After of this procedure we had a total of 1028 observations (i.e. 86.9%, 1028/1182).    

Bivariate analysis

We conducted the Chi-squared independence test between the AT device type provided (CodeATVD5C) and DiagCode7C, ImparCode3C, AgeCode4C, GenderCode, Caregiver, EcoStraCode3C, ZNCode3C, AfilTypeCode2C, YearCode3C, and NprovAT variables.

Then we obtained the Pawlik’s Contingency Coefficient between the AT device type provided (CodeATVD5C) and DiagCode7C, ImparCode3C, AgeCode4C, GenderCode Caregiver, EcoStraCode3C, ZNCode3C, AfilTypeCode2C, YearCode3C, and NprovAT variables. We also obtained the Cramér’s V association coefficient between CodeATVD5C and DiagCode7C, ImparCode3C, AgeCode4C, EcoStraCode3C, ZNCode3C, AfilTypeCode2C, YearCode3C, and NprovAT variables; and the Phi association coefficient between the AT device type provided CodeATVD5C and GenderCode and CareGiver. We included Pawlik’s Contingency Coefficient and Cramér’s V/Phi statistics to compare both results.

 

RESULTS

Univariate analyses

Table 1 illustrates descriptive data off all variables included in our study after the data imputation procedure. The most frequent type of AT provided was wheelchair (41.8%).  The two most frequent diagnoses were Diseases of the ear & mastoid process (17.5%) and cerebral palsy (16.9%). Motor impairment was the most frequent impairment (74.2%). Although the high frequency AT provision observed were under age (30.2%) and older adults (25%) age categories, frequency of provision by age did not differ more than 10% amongst categories.  This trend was similar for the case of gender (i.e. 44.1% of the overall clients were female, whereas 55.9% was male).

TABLE 1: Descriptive statistics2.

Independent variables

Operationalization of variables  

%

CodeATVD5C

 

0: Wheelchair

41.8%

1: AT for walking

18.4%

2: Ortheses and prostheses

13.8%

3: AT for hearing

17.3%

4: Other assistive products

8.7%

DiagCode7C

 

0: Diseases of the ear & mastoid process

17.5%

1: Cerebral palsy

16.9%

2: Diseases of the nervous system

13.8%

3: Diseases of the musculoskeletal system

15.2%

4: Injuries

12.3%

5: Cerebrovascular diseases

7.2%

6: Other diseases

17.1%

ImparCode3C

 

0: Motor impairment

74.2%

1: Hearing impairment

18.1%

2: Multiple impairment

7.7%

AgeCode4C

 

0: 0-18 years (Under age)

30.2%

1: 19-40 years (Young-middle adults)

20.6%

2: 41-65 years (Middle age adults)

24.2%

3: >66 years (Older adults)

25%

GenderCode

 

0: Female

44.1%

1: Male

55.9%

CareGiver

 

0: Not

18.2%

1: Yes

81.8%

EcoStraCode3C*

 

0: Rural

10.3%

1: Low social economic strata

65.2%

2: Medium social economic strata

24.5%

ZNCode3C

 

0: North

14.6%

1: South

70.1%

2: Downtown

15.3%

AfilTypeCode2C

 

0: Contributive**

20.4%

1: No-contributive***

79.6%

YearCode3C

 

0: 2002-2003

9.7%

1: 2004

35.4%

2: 2005

54.9%

NprovAT

 

1

82.6%

2

15.1%

3

2.3%

Endnotes

Number of data base observations: 1028

*: The Colombian system of economic stratification is a complex system based on the physical features person living setting. The system is not based on family income  (DANE, 1994)

**: Clients who pay for their insurance health coverage

***: Clients who are no able pay for their insurance health coverage. Therefore, the government assumes their basic health services. 

In contrast, differences in device provision frequencies according to the client`s socio economic strata, the geographic zone in which the client lives, the client`s affiliation type, the year of AT provision, and the total number of AT provided to each client were larger (see Table 1 for more details). Overall, the greatest proportions of clients receiving AT devices were located in urban areas of low social economic strata (65.2%), of the city south regions (70.1%), who belong to no-contributive healthcare service type (79.6%).

Also, it can be seen that frequency of AT provision increased from 9.7% in 2002-2003 to 54.9% in 2005. Finally, the frequency of clients who receive at least one AT device was very high (82.6%) in comparison with those clients who received 2 (15.1%) or 3 (2.3%) AT devices.

Bivariate analysis

Table 2 illustrates the Chi-squared independence test and association coefficients between the AT device type provided and all independent variables of our study.

TABLE 2: Chi-squared and association statistics.

Variables

c2

df

Stat.

Sig.

Pawlik CC

Cramér's V or Phi coefficient

DiagCode7C

1084.5

24

0.000

0.802

0.514

ImparCode3C

1003.1

8

0.000

0.861

0.699

AgeCode4C

156.79

12

0.000

0.420

0.225

GenderCode

4.16

4

0.385

0.089

0.064*

CareGiver

61.01

4

0.000

0.335

0.244*

EcoStraCode3C

134.10

8

0.000

0.416

0.255

ZNCode3C

75.36

8

0.000

0.320

0.191

AfilTypeCode2C

9.27

4

0.055

0.134

0.095

YearCode3C

87.06

8

0.000

0.342

0.206

NprovAT

154.27

8

0.000

0.442

0.274

Endnotes

Number of data base observations: 1028

Outcome variable: Assistive device type provided (CodeATVD5C)

c2: Chi-squared statistics

df: Degrees of freedom

Stat sig: p value (significant level at p<0.05)

CC: Contingency Coefficient (0, no association level, 1 variables completed associated)

*:  Phi coefficient (only for dichotomous independent variables )

The exploratory bivariate analyses showed that type of AT device provided was significantly dependent (p<0.001) of the diagnostic type, the type of impairment, the client`s age, whether the client has care giver, client’s socio economic strata, thegeographic zone in which the client lives, the year of the AT provision, and the total number of AT delivered. In contrast, the client’s gender and affiliation healthcare service type were not (p>0.05).

The Pawlik’s Contingency Coefficient showed that the diagnostic type (C=0.802) and the type of impairment (C=0.861) variables, have the most strongest association level with the type of AT device provided. In contrast, the client’s gender (C=0.089), whether the client has care giver (C=0.335), thegeographic zone in which the client lives (C0.320), the healthcare affiliation type (0.134), and the year of the AT provision (0.342) had weak levels of association. The Cramér’s V and Phi coefficients showed similar results than Pawlik’s Contingency Coefficient; nonetheless their values were lower than Pawlik’s Contingency Coefficient.

DISCUSSION

Our results allow us to give insights and explanations about the AT provision in a low-income country. First, the most frequent type of AT device provided were wheelchairs and AT for walking. This could be explained by the fact that, based on the clients´ diagnoses frequencies into the database, most of the (i.e. about 65%, see Table 1) diseases associated to the diagnoses are expected to generate a motor impairment, which had the highest frequency into the impairment category. Additionally, the type of AT provided showed the strongest coefficients of association with the diagnosis type and the type of impairment factors. One possible explanation for this result is that the provision process was based on the ICIDH. Thus it is probable that the client´s body structures and functions were the most important criteria during the AT prescription process.

Our results are consistent with the public policy for disability at the city, which prioritizes people who are in the highest levels of vulnerability. First, most of the AT devices were provided to clients who presented the highest levels of vulnerability. This is, clients’ at the lowest social economic strata who tended to live in the city´s south regions and belonged to the no-contributive healthcare service affiliation type. Second, under age clients and older adults received more than half of the AT devices provided. The independence between the type of AT provided and the healthcare service type, might be owned to the fact that regardless the healthcare service affiliation type, the Colombian general healthcare system does not cover AT devices such as wheelchairs. As a result, people must access to the AT provision through the BAT service. Thus, it is probable that during the provision process, the social economic strata had more importance than the healthcare service affiliation type.

Similar to previous studies (Hubbard, et al., 2010), we found that the type of AT provided was not dependent on gender. In spite of the relationship between diagnosis and gender, we verifiedhomogeneous frequencies of the type of AT provided by gender (i.e. contingency tables  GenderCode & CodeATVD5C)2. The fact that the highest frequency on quantity of AT provided was only one AT device per person might be due to in lower-income countries, the need for AT exceeds availability (Jefferds, et al., 2010). This situation might push AT provision’s decision makers to prioritize coverage over the client´s needs.

Study Limitations

We faced the typical limitations of studies that use administrative databases (Hubbard, Fitzgerald, Reker, Boninger, Cooper, & Kazis, 2006). In our case, it was a cross-sectional study limited to 4-year period, we had missing data, the database had no information about the clients` functional status, AT usability; and finally, the data were based on the ICIHD instead of the ICF. Further versions of the database should be based on the ICF.

CONCLUSION

The results of this study allowed us to conclude that: first, the type of AT provided depended on the diagnostic type, the type of impairment, the client’s age, whether the client has care giver, client’s socio economic strata, and thegeographic zone in which the client lives. Second, the diagnostic type and the type of impairment had the strongest association level with the type of AT device provided. Therefore, all the stakeholders who provide AT should mainly take into account these factors for developing public policy and clinical actions (e.g. client’s evaluation process) regarding the AT provision in Bogota.

REFERENCES

Barnard, J., & Meng, X. (1999). Applications of multiple imputations in medical studies: from AIDS to NHANES. Stat Methods Med Res , 8, 17-36.

Bernd, T., Van Der Pijl, D., & De Witte, L. (2009). Existing models and instruments for the selection of assistive technology in rehabilitation practice. Scandinavian Journal of Occupational Therapy , 16, 146-158.

DANE. (1994). Colombian socioeconomic stratification law. Consulted: January de 2013, de http://www.dane.gov.co/index.php?option=com_content&view=article&id=366&Itemid=114

Finch, H. W. (2010). Imputation Methods for Missing Categorical Questionnaire Data: A Comparison of Approaches, Ball State University. Journal of Data Science , 8, 361-378.

Hubbard, S., Fitzgerald, S., Reker, D., Boninger, M., Cooper, R., & Kazis, L. (2006). Demographic characterisitics of veterans who received wheelchairs and scooters from Veterans Health Administration. Journal of Rehabilitation Research & Development , 43 (7), 831-844.

Hubbard, S., Cowper, D., Wu, S., Reker, D., Vogel, B., Fitzgerald, SG; Mann, W; Hoenig, H.  (2010). Demographic and clinical variation in veterans health administration provision of assistive technology devices to veterans poststroke. Arch Phys Med Rehabil , 91, 369-377.

ISO/FDIS. (2006). Assistive products for persons with disability- Classification and terminology. Consulted:January de 2013, de http://infostore.saiglobal.com/store/Details.aspx?productID=411667

Jefferds, A., Beyene, N., Upadhyay, N., Shoker, P., Pearlman, J., Cooper, R., Cooper, RA and Wee, J. (2010). Current State of MobilityTechnology Provision in Less- Resourced Countries. Phys Med Rehabil Clin , 21, 221–242.

Karmarkar, A., Diciano, B., Graham, J., Cooper, R., A, K., & Cooper, R. (2012). Factors associated with provision of wheelchairs in older adults. Assistive Technology , 24, 155-167.

WHO. (2010). International Diseases Classification (ICD-10) version 2010. Consulted: January de 2013, de http://www.who.int/classifications/icd/en/

WHO. (1980). The International Classification of Impairments, Disabilities and Handicaps (ICIDH). Geneva: WHO.

ACKNOWLEDGEMENT

We would like to thank Research Grant of Universidad del Rosario for the resources provided in Grant DVG 142 FIUR/2012, which financed the study.     

Footnotes

  1. In spite of the most recent model of disability used is the international classification of functioning, disability and health (ICF) we did not use it owing to this information was not available in our data source.
  2. Contingency tables amongst CodeATVD5C and independent variables were not reported owing to space constraints. 

Audio Version PDF Version