Tools for data collection, analysis and application in weight loss clinical databases
**Communicating with other doctors**
A computerized database can automatically generate reports, facilitating efficient communication of a patient's condition with their family doctor or other medical professionals. We can typically use this database to automatically generate summary reports of a patient's condition and deliver them regularly to the patient's family physician.
**Conduct research, data verification, and quality control.**
A proper clinical study requires three things: ① a sufficient number of patients; ② accurate parameter measurements; and ③ maintaining data that is easily accessible and meets analytical requirements. A comprehensive and accurate database is the lifeblood of clinical research. Because we have accumulated a sufficient number of patients, we are able to publish a large number of articles on the clinical effects of gastric banding surgery. By acquiring and recording patient data and storing it in a database, this data can be linked to all clinical questions.
Not every healthcare professional needs to conduct clinical research, but they should all be involved in clinical data verification and quality control. All bariatric surgeons must establish and maintain a database of patient data to check whether patient outcomes are meeting expectations. For example, how much weight was lost? How many patients were lost to follow-up? What were the mortality or perioperative complication rates? What was the repeat surgery rate? How do their surgical outcomes compare to those of their colleagues? These questions can all be answered if a suitable database is established.
**Data and Analysis**
The scope of collectable clinical data is virtually limitless. There are two common drawbacks to establishing clinical databases: first, attempting to collect all information because someone might be interested in it someday; and second, treating the database as a supplement to medical records rather than an integral part of them. Collecting too much information is nearly impossible, as staff have other commitments. If a clinical database is independent of patients' daily treatment and care, it won't receive priority from staff. Therefore, every piece of data should have a reason for being stored.
We have been using electronic databases in bariatric surgery clinics for many years. The following items showcase our database, called LapBase, which currently collects clinical data covering the following ranges and can automatically calculate other parameters using software:
Demographic information: Name; Address; Home phone number, work phone number, and mobile phone number; Email address; Gender; Date of birth; Name and contact information of family doctor; Name and contact information of specialist doctor.
Anthropometry data: weight; height (BMI, ideal weight, target weight); neck circumference, waist circumference, hip circumference; blood pressure; total body fat mass, fat percentage; patient photograph.
Accompanying diseases: hypertension, diabetes, dyslipidemia, asthma, gastroesophageal reflux, urinary incontinence, sleep apnea, infertility, lower back and joint pain, heart disease, and other diseases.
Surgery and other important details: type of surgery, including important variables, such as Roux loop length and type of gastrojejunostomy for RYGB; date of surgery; surgeon, operation time, and length of hospital stay; complications; postoperative barium meal examination.
Treatment outcomes: weight, health status, quality of life, amount of fluid injected into the laparoscopic adjustable gastric banding (LAGB) band, and annual comorbidity screening.
Medical records should be documented in text form. Each barium meal examination should be recorded in the follow-up data.
**Common Methods for Clinical Data Databases**
Beyond displaying the input of clinical data, the output of analytical results is determined by the data management methodology. This ranges from simple text input and traditional backups of medical records to comprehensive databases that allow for cross-analysis across all domains.
**text**
Create a file for each patient. Input all patient information as text; the result is a medical record, very similar to the paper medical records we are all familiar with. As long as its structure is logical, all parts of the record can be accessed. Different patient files can be accessed using alphanumeric codes. Clinical data cannot be statistically analyzed without categorizing and inputting it into tables or databases. Clinical data should not be recorded like paper medical records; instead, it should be categorized and organized.
**spreadsheet**
Spreadsheets are often described as flat databases; they are relatively simple systems, ideal for storing basic information. A typical spreadsheet has each column containing a specific parameter for all patients, while each row contains all clinical parameters for a particular patient. By establishing such a data structure, data can be analyzed within a column, special manifestations can be categorized, specific parameters can be identified, and a series of mathematical and statistical analyses can be performed on individual cells or columns. However, the analysis capabilities of a spreadsheet are limited by the content of the initially created rows and columns.