Table 9 Reading data set into the R system

ID

Coded number for each patient

Each patient has a unique code

This will only be used to produced graphs where needed

Name

Name of patient

Those who refuse to allow the use of their true name will be replaced by N.A. Some patients may have identical names, and they will be distinguished by ID

This will not be used

Res

This answers if the patient is resident of the region where the hospital is located

This takes binary value of yes or no

This has a potential value for implying transports and related costs for patient’s relatives

Days

Number of days the patient spent in the hospital

Quantitative variable

This is for charting the frequency distribution and transforming into long or short stay

Stay

A dummy variable to define if length of stay in hospital by a patient is short or longer

This takes value of S if one’s stay is less than 10 days, and L (longer) if 10 days or longer

This is potentially a good candidate for a binary predictor variable

Insured

Whether a patient has a valid health insurance policy

This variable is binary and takes value of “Yes” or “No”

This is potentially a good candidate for a binary predictor variable

MaxIns

The maximum level of coverage by insurance policy if the patient has one

Quantitative factor, measured by percentage of total “eligible costs” according to regulations

This is supplementary information only

Edu

Highest level of education

Values:

JHS: junior high

HS: high school

Uni: college/university

Grad: graduate school

This is supplementary information only

SES

Socio-economic status

This is a multicategory variable, taking values of: high, medium, low

This is potentially a good candidate for a multi-categorical predictor variable

Illness

Degree of severity of illness or injury when hospitalized

This is a multicategory variable, taking values of: emergency, bad, ill and light

This is potentially a good candidate for a multi-categorical predictor variable. Ill and Light categories can also be grouped into one single category, in which case we have a transformed factor of Ill2 (also in the data set)

Jcond

Status of job

This takes value of good, stable, unstable, unemployed or N.A. (i.e. others: retired, high school students)

This is supplementary information only

Income

Annual income in millions of Vietnamese Dong. Current exchange rate: $1 = VND21,200 (as of Dec 1, 2014)

Quantitative factor. The factor can be continuous in theory, but is mostly categorical in practice

This will be used to derive the IncRank variable

IncRank

Rankings of income of a patient

This takes value of high, middle, or low

This is potentially a good candidate for a multi-categorical predictor variable

Spent

Total money spent during his/her stay in hospital in millions of Vietnamese Dong. Using official exchange rate, VND 1 million is equivalent to $47.2

Quantitative variable

Dcost

Average daily cost the patient had to pay during treatment period

Quantitative variable

Pins; Pinc; Pchar; Ploan

Portions of finance from sources: insurance reimbursement, income, charity funds from civil organizations or employers, or borrowings

Quantitative variables. They are measure in percentage of total costs the patient had to cover

InsL2

Ranked levels of actual insurance reimbursement for patients who have policies

It takes value of high, medium, or low if a patient has an insurance policy. It takes nil if the patient is not insured at all

For actual transformation when necessary both low and nil can be grouped into one category of Neg (i.e., negligible)

Streat, Srel, Senv

Percentage of funds used for the purpose of main treatments, for covering costs of relatives coming to help the patient, or paying “extra thank-you money” or bribing doctors/staff

These are quantitative variables, taking value of percentage. For instance, patient ID001 has {89;4;7 %} = {0.89; 0.04; 0.07}. Streat + Srel + Senv = 100 %

Only ‘Senv’ can be a potential candidate for modeling as dependent categorical variable, after being transformed into levels of “extra fees”, taking value of high, medium, or low. that newly derived categorical variable is called EnvL and also appears in the data set

Burden

Patient’s and family’s self-evaluation of their financial position after paying health care costs

This is a multi-category variable, taking value of A (strong; no adverse affect at all); B (affected but not the worrying level); C (seriously affected) and D (destitute/“bankrupt”

This represents a group of critically important categories that the study employs to learn about what factors likely affect the probabilities of falling into each category of financial burden

End

The health status after treatment

This is a group of multi-categorical variables, taking values of A (complete recovery), B (partial recovery, needing post-treatment follow-ups), C (stopped whilst being treated), or D (quit early)

This represents a group of critically important categories that the study wants to model to know what factors affect the probabilities of falling into each category of treatment completion after patients’ stay in hospital

IfHigher

Self-evaluation of patient and family about financial status if the patient continues to be hospitalised again

This takes the same values as “Burden” factor

This can potentially be a candidate for future examination, especially when a larger sample is available. It is not used for this analysis