Health exams in Vietnam

surveys
classification
logistic regression
As access to health care increases, public health advocates encourage ordinary people to get regular check-ups to detect problems before they become serious (and expensive). But not everyone follows this advice. This survey of people in Vietnam explores the factors leading people to get regular check-ups.
Author

Alex Reinhart

Published

December 8, 2022

Data files
Data year

2017

Motivation

Medical care for serious diseases can be very expensive, and in countries without established socialized medical systems it can impose serious hardships on patients. Ideally, patients would get regular check-ups (or “general health examinations”, GHEs) so that serious conditions could be detected and treated early, before they cause serious problems. This would save money and improve public health.

However, there are many possible obstacles to getting everyone to go to regular check-ups. They might be too expensive, or too difficult to schedule; some people may not trust doctors or believe that check-ups have any value; or some people may have had bad experiences when they previously tried to get check-ups.

Public health researchers in Vietnam wanted to explore these reasons and determine what obstacles prevented widespread use of regular check-ups. They conducted an interview survey in Hanoi and Hung Yen, Vietnam, by traveling to “secondary schools, hospitals, companies, government agencies and randomly selected households in Hanoi” and interviewing people in person for about 10-15 minutes. This dataset contains the raw data from that survey.

Data

There were 2,068 valid responses to the survey, out of 2,479 people approached. This dataset includes all the valid responses.

Data preview

vietnam-health.csv

Variable descriptions

Variable Description
id A unique ID for each survey respondent
date Probably the date of the interview, in YYYYMMDD format
Age Age of survey respondent (years)
Sex Sex of survey respondent
Jobstt Job status: stable, unstable, student, retired, homemaker, or other
MaritalStt Marital status: married, unmarried, other
Edu Maximum educational attainment (PostGrad = post-graduate, Grad = college or university, Second = high school, Hi = middle school)
height Height (cm)
weight Weight (kg)
place Place the survey was conducted
BMI Body mass index, calculated from height and weight
HealthIns Whether the respondent has health insurance
RecExam Time since the respondent last visited a doctor with symptoms of a disease (less12 = less than 12 months, b1224 = between 12 and 24 months, g24 = over 24 months, unknow [sic] = respondent doesn’t know)
RecPerExam Time since respondent last visited a doctor for a check-up, not prompted by a specific illness. Coded in the same way
ReaExam Reason for their last check-up (noti.disease = concern over illness or epidemic, adv.sig = worrying symptoms, request = requested by an employer or insurance company, volunteer = no immediate reason)
Wsttime Does the respondent believe check-ups are a waste of time?
Wstmon Does the respondent believe check-ups are a waste of money?
DiscDisease Is the respondent afraid of check-ups because they’re afraid of discovering a disease?
Lessbelqual Does the respondent have little faith in the quality of medical service?
NotImp Does the respondent believe check-ups are not urgent or important?
HthyPriority Does the respondent believe health is a first priority?
ComSubsidy Are check-ups subsidized by the respondent’s employer or community?
Habit Did the respondent get the habit of regular check-ups from their family or employer?
FlwHealth Does the respondent constantly follow updates on their health measures?
PerTrmt Is the respondent receiving long-term medical treatment?
AcqTrmt Is a member of the respondent’s family receiving long-term medical treatment?
StabHthStt Are the respondent and their family all in good health?
MedCabinet Does the respondent keep a medical cabinet and basic medical equipment at home?
Tooluseskills Does the respondent have the skills to use basic medical equipment?
ExpCare Does the respondent have experience taking care of a sick family member?
ExamTools Does the respondent’s family regularly take simple medical measurements, like blood pressure or weight?
UseMon If the respondent were provided cash to have a check-up, what would they do? allsoon = use all the money to have a check-up soon, partly = use part of the money for a check-up and save the rest, later = take all the money and have a check-up later
Tangibles Perceived quality of medical equipment and personnel at check-ups. Scale of 1 to 5, 1=lowest quality
Reliability Perceived ability of examiner to perform medical services that meet the patient’s expertations, scale of 1 to 5
Respon Perceived timeliness of check-up service, scale of 1 to 5
Assurance Perceived knowledge or ability to assure professional reliance, scale of 1 to 5
Empathy Perceived thoughtfulness and sense of responsibility of medical staff, scale of 1 to 5
StChoise Respondent’s preferred way of dealing with new symptoms: clinic = go to a clinic, askrel = ask family and relatives for advice, selfstudy = do personal research.
CHPerc Respondent’s general opinion of public health (good, quite good, bad, unknown)
SuitFreq How often respondent believes check-ups should be done (6m = every 6 months, 12m = every 12 months, 18m = every 18 months, g18m = less often than every 18 months)
AffCost How much the respondent believes a check-up should cost (low = less than 1 million VND, med = from 1 to 2 million VND, hi = over 2 million VND)
UseIT Is the respondent willing to use information technology to detect health problems, if it’s reliable?
AfterIT If a healthcare app indicated that the respondent needed to have a check-up, would they arrange one?
SuffInfo Respondent’s rating of the sufficiency of information they received in check-ups, 1 to 5 scale
AttractInfo Respondent’s rating of the attractiveness of information they received in check-ups, 1 to 5 scale
ImpressInfo Respondent’s rating of the impressiveness of information they received in check-ups, 1 to 5 scale
PopularInfo Respondent’s rating of the popularity of information they received in check-ups, 1 to 5 scale
SuitExer How much time the respondent thinks people should spend on sports and physical exercise to stay in shape (units not given)
EvalExer How much time the respondent spends on sports and physical exercise (verysuff = ‘more than enough’, quitesuff = ‘enough’, little = ‘only a little’, trivial = ‘none or almost none’
Age_gr The respondent’s age, broken into one of five categories

Questions

Suppose the Assistant Minister of Health is interested in increasing the number of people who get annual health exams. The Ministry could simply send reminders or offer incentives to anyone who does not appear for an exam each year. But the budget is limited and they cannot afford to try to convince every possible patient to get a check-up. Also, the Ministry does not know why people do not sign up for exams, and understanding why could help them make policy changes or make their public relations campaigns more effective. (For example, if people tend to think check-ups are a waste of money, their advertising should emphasize how cheap they are; if people don’t believe medical service is high-quality, the advertising should emphasize the expertise and training of doctors.)

  1. Overall, how do people rate the attractiveness, impressiveness, sufficiency, and popularity of information they receive in checkups? Give summaries of these variables, as well as variables like assurance, reliability, and empathy that indicate how medical personnel may be able to improve.
  2. Build a classifier to predict whether the respondent had a check-up in the last 12 months (using RecPerExam). Use predictors that plausibly could be related, and discuss which predictors have the strongest relationship with the outcome. Which make a person less likely to get a check-up every twelve months?
  3. The original paper (see citation below) used logistic regression to evaluate how RecPerExam can be predicted by Wsttime, Wstmon, HthyPriority, FlwHealth, and HealthIns. Try to replicate their analysis and evaluate whether logistic regression is a good modeling choice. How do its results compare to your classifier?
  4. One plausible hypothesis is that none of these factors matter much unless you have health issues or know someone in their family with health issues. (This is in StabHthStt.) That is, if nobody you know has serious health issues, you may not decide to get check-ups, regardless of how attractive impressive or inexpensive they are. If you do, then the factors do influence your probability of getting a check-up. Test this hypothesis.

References

Q-H Vuong (2017). Survey data on Vietnamese propensity to attend periodic general health examinations. Scientific Data 4 170142. https://doi.org/10.1038/sdata.2017.142

Data available from the Open Science Framework: https://osf.io/afz2w/