This is a case–control study comparing cases with type 2 diabetes to controls without diabetes matched on age and sex. The case–control study design isolates the costs caused by diabetes by revealing differences and ratios between persons with diabetes and matched controls. The method solves the otherwise very difficult problem of allocating joint costs-for example costs for a heart attack admission that are partly caused by diabetes but also partly due to other causes.
Data will be obtained by personal interviews with the cases and controls. Interview data will be supplemented with data from health facilities to more fully describe the subjects’ costs of care. Searching of health facility records will be based on personal-interview responses that indicate when and where subjects have received medical care. The form for unit costs data collection from health facilities will be piloted in a similar setting in Bangladesh before the study. All subjects will provide informed consent prior to being interviewed and having their health facility records examined.
Sampling design
Case selection and recruitment
Cases with diagnosed diabetes will be selected from the Bangladesh Institute of Research on Diabetics, Endocrinology and Metabolism (BIRDEM) Hospital outpatient department (OPD) where patients from all over Bangladesh comes for management of diabetes and its complications. We will identify cases with diabetes from consecutive patients as they come for treatment. Potential subjects will be asked if they are willing to participate in the study and, if so, a time and location for a personal interview will be arranged and contact information gathered.
Selection and recruitment of control subjects
Controls will be individuals ‘frequency-matched’ to cases on age (within 5 years band) and sex (male or female), without a self-reported history of diabetes (or diagnosed as diabetes by a physician). They will be recruited within 48 hours of recruiting the index case, from either (i) the same geographical residence of cases, or (ii) visitors of patients attending the out-patient department, or (iii) non-blood related visitors of index diabetes cases, in the same hospitals,. All controls will undergo identical study questioning and examination as cases. One completed control interview will be obtained for each case interview.
While we are recruiting case subjects and scheduling personal interviews, we will ask cases who have said they want to participate to identify five persons living closest to him or her who are of the same age (+/− 5 years) and sex. For each of these potential control subjects, we will ask the case to provide whatever contact information that he or she can. Cases will be assured that their health information will not be shared with the control. We will then randomly order the names on the resulting list of five and approach each of these individual’s households in order, by telephone if possible or, if not, by visiting their households. If the first-ordered individual is at home, the interviewer will recruit this individual on the spot and obtain informed consent and complete the interview immediately if this is practicable. Otherwise, if the individual is not willing to be interviewed at the moment, an appointment for a later time will be made. If the first-ordered individual is at home but does not wish to participate, the interviewer will proceed to the household of the second-ordered individual and repeat the process.
If the second-ordered individual is not at home, the interviewer will attempt to get telephone contact information for this person so that recruitment can occur by phone at a later time, and an interview scheduled. Interviewers should not leave the area where they recruited the case without a lead to a potential control. If by the second household the interviewer has not recruited and interviewed a control, she/he will proceed to each of the remaining households to obtain contact information and gauge potential interest.
The approach described above is chosen because it can potentially achieve a balance between feasibility and scientific rigor and should be scalable in Bangladesh. By contrast, whereas use of population-based community controls may be desirable in principle, it is considerably more labour-intensive and expensive and cannot guarantee that such controls will necessarily represent the catchment areas from which hospital-based cases are derived, particularly since referral patterns to hospitals are complex. Furthermore, as controls in this study will be drawn from visitor of patients attending the out-patient department and non-blood related visitors of diabetes cases, it is expected that such approach would minimize, at least in part, major limitations involved with conventional hospital-based controls if we cannot recruit community based controls.
Use of surrogates
When a case or control subject is not able to speak or recall sufficiently completing the interview on their own, the participation of surrogate respondents will be allowed. The interviewer will ask another family member to answers the survey questions on behalf of the respondent. The name of the family member who answered the questions will be written in the survey and the surrogate code entered on the survey.
Non-response, non-completion, termination
If it proves impossible to complete an interview once it has begun, and it cannot be completed on another day, this case or control will be removed from the study sample and replaced with another person per protocol. They will also be labeled a “non-completer.” If a subject desires to terminate an interview prior to completion, the same procedures will be followed as for a non-completer. All data except for recruitment information obtained prior to commencing the interview will be destroyed. The interviewer will terminate the interviews of all control subjects who say that they have been diagnosed with diabetes, or whenever in the interviewer’s judgment the subject is not responding truthfully or accurately or no longer wishes to participate in the study.
The research tools and instruments will include pre-defined questionnaires, anthropometric measurements, laboratory and medication lists, and physical examination. We will use the research tools developed by the International Diabetes Federation (IDF) Health Economic Group and translate them into Bengali as per WHO standards. All interviews will be conducted by trained interviewers under the supervision of a Research Physicians.
Hospital data gathering tools: Information on cost incurred by the hospitals is a vital component of illness costs. Information on treatments received by patients during their hospital visits is also critical in understanding utilization of health care. To generate cost incurred by third parties and to gather hospital information, we have developed a hospital unit costs tool to gather information on average costs including those incurred by third parties (by type and reference level of the hospital or outpatient clinic).
Estimation of expenditures for medical care will be based on participant recall to ascertain charges for medicines, supplies, and medical care services. To increase accuracy of recall, the interview schedule will only ask about events occurring only during the previous 90 days and attempt to improve temporal accuracy by asking respondents to name and associate a well-remembered event that had occurred approximately 90 days previously. To estimate expenditures for medicines, interviewers will ask subjects about their most recent purchase of each of the medicines they were currently using. For overnight admissions to hospital and OPDs, respondents will be asked to recall their total point-of-service payment, including charges for medicines and tests received or, if they paid only a portion out of pocket, the total bill or charge. Total charges per visit or hospital admission usually includes charges for medicines because doctors and hospitals typically use pharmaceutical sales as a primary means to finance operations. This subset will be used to estimate the characteristics of all events of the same kind, including mean length of hospital stay and mean payments per admission, per OPD visits, and per purchase of medicine.
Field testing
The research instruments and questionnaires and will be translated in Bengali and field tested in a similar setting to our current study site and population like the outpatient department of Bangladesh University of Health Science (BUHS) Hospital, Mirpur, Dhaka before conducting the research. 25 cases and 25 control subjects will be enrolled in the field testing. Feedback from the field testing will be used to improve the language and contents of the questionnaire and tools.
General outline plan for field work/data collection
On a typical day, the field staff will be present at the study site early in the morning when patient registration starts in the BIRDEM OPD. Registered patients who agree to participate in the study will be referred by the physician on duty to the study team. The study team will explain about the study in details to the participants and request for signing the informed consent to all who agree to participate in the study on a volunteer basis. The study team will also collect data from one control against one case.
On an average, 5–6 patient forms should be completed at the study site every day and the total data collection should be completed by 12 months. At the end of the day all team members will ensure that the completed data collection forms and instruments are properly stored in the field site. Data will be entered in an excel sheet in an ongoing basis. A log book of all participants with codes will be maintained by the study team. Written informed consent will be obtained from all participants that they agree to participate in the study voluntarily and are free to stop participating at any point of the study.
Follow-up plans
Quality Monitoring and Assurance: Data collection forms will be checked on a daily basis to ensure complete and accurate data entry by field research workers. Cross checking of data collection forms will be performed in the fields under supervision of the Research Officer for inconsistency and missing data. The Principal Investigator (PI) will be responsible for carrying out periodic data checks and he will look for systematic patterns, errors, scheduling problems or incomplete forms to provide feedback to the study nurse and to ensure integrity of data captured. Every week when the team meets, the PI will pick 2 questionnaires filled in the current week by one interviewer. The PI will go through the questionnaire, question by question with the entire team, to identify incomplete/missing entries and any errors. The purpose of this is to rectify and clear any doubts which the interviewers may still have.
The PI will visit the fields at least once every week and observe complete data collection by the field research assistants. He will provide his feedback and provide on-the-job training as necessary. Co-PIs and Co-Investigators will also visit the field site without prior notice to monitor data collection, physical measurements and biological sample collection, transportation, storage and analysis in accordance with the protocol. The instruments for physical measurements will be regularly calibrated according to guidelines of its manufacturers in the fields and a written record maintained which will be periodically checked by the PI for consistency and validity of data.
We will attempt to replicate interview and physical measurements in a random sub-set (~5%) of the sample to evaluate repeatability. For repeatability studies, we will have about 50 participants, 25 cases and 25 controls, with duplicate measures which will be sufficient to estimate kappa coefficients of agreement of inter and intra-observer reliability with a high degree of precision even for categorical variables. The participants will be selected among study participants who come to BIRDEM OPD for treatment and follow up and agree for the process.
Description of variables
-
Socio-demographic and economic information: We will collect data on participant’s age, sex, religion, education, marital status, occupation, income, family size, family income, objects owned
-
Knowledge of prevention, management and complications of DM: Levels of knowledge and perceptions of diabetics, Risk Factors for diabetics, knowledge about prevention, management and complications of diabetics.
-
Self-perceived Quality of Life (OoL): This will be measured using EQ-5D 3 L, which is a standardised instrument for use as a measure for health outcome. It is widely used, and by many governments a recommended instrument for measuring health related QoL [
18].
-
Self-reported diseases: This will be based on self -reported information on diabetes, cardiovascular diseases, renal disease, stroke, arthritis, chronic respiratory disease, mental and physical disability by the study participants.
-
Medical History: Any known history of diabetes and its type, CVD, hypertension, cancer, hypercholesterolemia, stroke, kidney diseases, other vascular disease, infectious disease, major surgery.
-
Tobacco Use (WHO modified): A modified WHO questionnaire on use of different forms of tobacco currently and in the past will be recorded. Expenditure on tobacco use will also be recorded [
19].
-
Physical Activity (GPAQ): WHO GPAQ questionnaire which has been validated in the fields will be used to capture information on different forms of physical activity in MET which will be analyzed as per data analysis protocol [
20,
21].
-
Diet: A modified WHO STEPS questionnaire will be administrated to ascertain food consumption pattern and dietary choices, type and frequency of consumption of different foods.
-
Health seeking behavior: Data will be collected on type of health care practitioner visited during the past year, number of visits, distance travelled for consultation and expenditure for health.
-
Use of medical services: Hospital and non hospital visits for health problems, medical tests performed.
-
Impact of health problems: affecting personal and family life.
-
Diabetes history: description of diagnosis, duration, self-management of diabetes.
-
Mental health status (Patient Health Questionnaire, PHQ9): The PHQ 9 which is a standard toll for collecting mental health status in primary health care setting will be used in this study[
22,
23].
-
Life events and Disability: Life events related questions like new job, marriage, separation, injury, accident in the family will be asked, which will be based on validated questionnaire.
-
Anthropometric Measurements: Weight, Height, Waist and Hip circumference, Blood Pressure (BP), Pulse rate will be measured.
-
Laboratory data: data on HbA1c, FBS, 2HAFB, S. Creatinine, Lipid Profile, ECG will be collected from patient record book as available.
-
Physical examination: General examination of eyes, foot and neurology tests performed by a specialist will be recorded.
Details about physical measurement
Measurement of blood pressure and pulse rate
Blood pressure is taken to determine the prevalence of raised blood pressure in the population. We will use Omron SEM-1 (Omron, Matsusaka Co., Japan) digital blood pressure measurement instrument which is widely used for research purpose in the community. Both systolic and diastolic blood pressure will be measured for each participant. Three repeated measurements will be recorded after an interval of 5 minutes alternating right and left hand.
Height
Height will be measured for all participants using portable stadiometer. Standing height is measured with the subject in bare feet, Back Square against the wall and eyes looking straight ahead. A set square resting on the scalp and a tape measurement from the wall/bed is used to measure height to the nearest 0.5 cm.
Weight
The weight of eligible participants will be taken to help calculate their body mass index (BMI), which is their weight relative to their height, and therefore to determine the prevalence of overweight and obese people in the population. Weight is measured in undergarments using a digital platform scale, to the nearest 100 grams.
Waist circumference
Waist circumference is measured to the nearest 0.1 cm using a non-stretchable standard tape measure attached to a spring balance exerting a force of 750 gm. Take measurement over the unclothed abdomen at the smallest diameter between the costal margin and the iliac crest. The tape measure must be kept horizontal for standing measurement. Subject should relax with arms held loosely at sides.
Hip circumference
Hip circumference is measured to the nearest 0.1 cm using a non-stretchable standard tape measure attached to a spring balance exerting a force of 750 gm. Take measurement over light clothing at the level of the greater trochanters (usually the widest diameter around the buttocks). The tape measure must be kept horizontal for standing measurement.
Data analysis
All data forms and questionnaires will be checked for errors and necessary corrections will be made before data entry. Data will be entered using data entry programme with built in range and consistency checks. Frequency distributions will be run to identify outliers. In all analysis, potential confounding variables and effect modifiers will be considered. Descriptive analysis will be performed for all variables and unadjusted comparisons between case and control will be made using T-tests (for continuous variables) or Chi-square Tests (for discrete variables). When appropriate, longitudinal models will be created for outcome variables. All data will be presented before and after adjustment for confounding and interaction. Basic presentations of data will include number and percentages of cases with diabetes by age, sex, location of residence, and in relation to different known risk factors. Chi-square tests will be used to assess the association between diabetes and different risk factors. Prevalence odd ratios will be calculated for diabetes adjusted for major confounding variables (e.g. age, sex, smoking habits, socioeconomic status, nutritional status, and family history of diabetes). Interaction between age and risk factors, obesity and diabetes risk factors, and nutritional status and risk factor (hypertension and diabetes risk factors) will also be examined in the logistic regression model. All statistical analyses will be done using SPSS (Version 20 from SPSS corporation, USA).
Estimation of expenditures for medical care: The mean length of hospital stay and mean payments per admission, per Out Patient Department (OPD) visits, and per purchase of medicine will be calculated. Calculation of annual medical expenditures: Annual rates of use of medical services will be calculated by multiplying the amounts self-reported for the preceding 90 days by 4. We will calculate 90-day OPD expenditures separately for visits made by a participant to hospital clinics, to the private offices of doctors, to pharmacies, and to community health workers.
We will calculate an average daily price, using data about the price paid the most recent time the item was purchased, the number of pills or units of insulin purchased at that time, and the number of pills or units prescribed per day. We will then multiply this result by an adjuster for self-reported adherence, the average number of days per week that the participant indicated that he or she adhered to the prescribed regimen for a given medicine, divided by seven to get the payment per day “as used.” Mean daily payments will be multiplied by 30 to obtain a monthly mean expenditure and by 365 to obtain an annual mean expenditure. Payments for glucose testing strips will be calculated similarly to payments for medicines except that self-reported testing rates (times per day x days per week or month) to be used in lieu of prescribed usage rates and adherence to obtain mean daily, monthly, and annual usage and expenditure.
We will calculate total expenditures for medicalcare per person as the sum of estimated annual payments per person for inpatient hospital admissions, annual payments per person for OPD, annual payments per person for medicines, and annual payments per person for glucose-testing supplies. To avoid double counting, because self-reported payments for OPDs and admissions included payments for medicines, we will subtract from the grand total payments for medicines and strips that were purchased from hospitals during visits and admissions. Based on patient self-report, we will calculate the proportion of hospital, clinic, OPD visits during which medicines were purchased.
Hypothesis Testing: We will use a two-step “hurdle” model to test for differences between persons with DM and persons without DM. Using multivariable logistic regression models that included case status (with DM =1, with Non-DM =0), the team first test for differences in the proportion of subjects with a non-zero value, e.g., persons with any OPD visits during the preceding 90 days, controlling for continuous age, sex, and urban vs. rural residence as well as case. Age will be entered as a linear continuous variable because the addition of other transformations on age, e.g., age squared, will not significantly improve the performance of the model. Then, a second identically specified multivariable model on the non-zero values. The functional form of the second regression model depend on the underlying distribution of values: for counts of admissions, OPDs, and medicines, a Poisson regression model will be used, while for average length of hospital stay and costs per event, we will use ordinary least squares. The hurdle approach does not yield a single overall coefficient or confidence interval for hypothesis testing. However, if the coefficient on DM status is positive and significantly different from zero in one model and positive and not significantly different from zero in the other model, or if both models are significantly positive for DM, then the two models together may be considered statistically significant. As our primary hypothesis is one-sided, we will use one-sided tests to assess significance.
Healthcare use and expenditures: Absolute results for health care use and costs associated with DM can be estimated by subtracting age-, sex-, and location adjusted estimates for persons with Non-DM from the relevant unadjusted results for persons with DM. However, for purposes of comparing results from this study to the results of other studies conducted in other places and times, a more robust statistic is the adjusted ratio of costs or use among persons with DM to those among persons with Non-DM (DM: Non-DM ratio). Ratios are much less influenced than absolute differences by variations in source data, recruitment bias, differences in economic systems, conditions and patterns of medical practice, and currency fluctuations. DM: Non-DM ratios will be calculated. Analysis of use of effective diabetes care: Estimates of the percentage use of essential medicines and other percentage measures of quality and access to medical care will be calculated only for persons with DM. Associations with categories of age and length of time since diagnosis of DM will be tested for statistical significance using multivariable logistic regression models with continuous age, sex, urban vs. rural residence, and continuous duration of DM as predictors.