We analyzed data were used from three clinical databases: Focus On Therapeutic Outcomes, Inc. (USA), Maccabi Healthcare Services (Israel), and the National Information Service for Allied Health Care (the Netherlands). Data were selected for patients aged 18 years or older who started an episode of physical therapy care between January 1st and December 31st 2005. This study was approved by the Institutional Review Board for the Protection of Human Subjects of FOTO and Maccabi. In the Netherlands, ethical approval was not obliged as patients were not subjected to treatment other than usual, nor were required to behave in a certain manner.
Clinical databases
Focus On Therapeutic Outcomes, Inc. (FOTO) is a proprietary international medical-rehabilitation data management company from the United States that has been in existence since 1992 [6, 11]. The FOTO network was developed for the purpose of generating an outcome-oriented, standardized information management system for use in outpatient physical therapy settings [12]. The company's purpose has been defined as: to provide reliable, valid and responsive outcomes measures and aggregate data management services to enable real-time information that empowers clinicians, patients, payers, and policy makers, and facilitates choice, delivery and payment based on the most effective rehabilitation therapy. In the current study, data of 1,004 physical therapists, working in 187 outpatient practices in 28 different states (U.S.) were used. More than 60% of outcomes data were entered via computer software employing computerized adaptive testing (CAT) methods [13–16], but paper and pencil data entry were available for clinics without computer availability. FOTO is the largest CAT generated outcomes data collection process for outpatient therapy in the world with over 2.4 million patient episodes and 700,000 CATs administered as of December 2007. Outcomes data are supplemented by process information and used by therapists to manage their patients in real time. Administrators use the data to manage the clinics and clinicians.
Maccabi Healthcare Services (Maccabi) is the second largest public Healthcare plan in Israel. Maccabi collects physical therapy data from over 70 outpatient clinics using several parallel informatics systems, which makes Maccabi the first health care service internationally to fully integrate electronic functional status outcomes assessment with an electronic medical record [17]: 1) electronic central medical file system; 2) electronic appointment management system; 3) central computer with the ability of querying the first two systems; and 4) computerized adaptive testing for functional outcomes measurement and data collecting. In the current study, data from 73 physical therapy clinics including over 400 therapists were used. Therapists use the outcomes, process and administrative data to manage their patients in real time, and both clinicians and physical therapy service managers use the data to improve patient management.
The National Information Service for Allied Health Care (LiPZ) is a computerized registration network in which about 100 Dutch physical therapists working in outpatient practices participate [6, 18–20]. LiPZ was implemented in order to provide up-to-date information about the care provided by allied health care professionals in the Netherlands. LiPZ has been collecting health care related information since 2001. Participants use computer software to register their patients and treatments. In this software a special LiPZ-application is included, making it possible to register additional data and to make an export file every month. The data contain demographic information about patients visiting physical therapists, as well as information about the patient's condition and subsequent treatments. In the current study, data from 94 physical therapists, working in 43 practices were analyzed. LiPZ data are used for research purposes and administrators can use benchmark data to manage the clinics.
Data set
None of the three data sets collected precisely the same information. However, there were similarities in data elements among all three databases, as the collection of data on patients' date of birth and gender, and on the profession of referring physicians. Furthermore, in all databases, the number of visits per episode, i.e. the number of times the patient had a face-to-face patient-therapist encounter, was collected. Data that needed recoding, because of differences between the datasets, were symptom episode duration, the patients' complaints and interventions. The recoding procedures are explained in the following paragraphs. These procedures were based on choices established on the basis of a consensus procedure among the authors.
Data on episode duration of the health problem, defined as the number of days between the date of onset of the condition and the date of therapy initial evaluation, was collected in all three networks as well, but the codes varied. In FOTO, the data were coded as '0 – 7 days', '8 – 14 days', '15 – 21 days', '22 – 90 days', '91 days – 6 months', '> 6 months'. In Maccabi, the data were coded as '0 – 21 days', '21 – 90 days', '> 90 days'. In LiPZ, the categories were '0 – 2 days', '3 – 7 days', '1 week – 1 month', '1 – 3 months', '3 – 6 months', '6 months – 1 year', '1 – 2 years', '> 2 years'. For our purposes, episode duration was recoded in 'acute' (less than 3 weeks in FOTO and Maccabi, less than 1 month in LiPZ), chronic (more than 3 months or 90 days) and sub acute (the category in between).
In all databases, information about the patients' complaints, e.g. reason for treatment, was collected. However, different classifications, with different levels of detail were used. As in all classifications the body part treated could be deducted, this was used as indication of the health problem of the patient. In FOTO, the patient, the front office staff or the therapist could select the body part treated. In Maccabi, the primary physical therapists' diagnoses were collected, using ICD-9 [21]. For the current study, these ICD-9 codes were recoded into the body part treated. In LiPZ, the reasons for referral as given by letter by the referring physician were coded by researchers using the International Classification of Primary Care (ICPC) [22]. These ICPC-codes were also recoded into the body part treated. Additional file 1 provides an overview of the response options in each database and the way they were summarized into the body part treated.
Interventions were collected in all databases, but time span of registration and classification differed. In FOTO, entry of interventions was optional for the therapist. When entered, each intervention is recorded for being applied at least once in the treatment episode or not at all. In Maccabi, the registration of intervention codes during the episode of care is mandatory, therefore the number of times each code was used during the overall episode of care is known. In LiPZ, at most three interventions applied in at least half of the treatment visits are registered at the end of the treatment episode. The different classifications are summarized into the following categories, deducted from the American Physical Therapy Association's (APTA) Guide to Physical Therapist Practice [23]: therapeutic exercises; functional training in work; manual therapy techniques; prescription, application, fabrication of devices; electrotherapeutic modalities; physical agents and mechanical modalities; and other. Additional file 2 gives for each database an overview of the response options and how they were summarized into the APTA categories.
The selection of patients with lumbar syndromes was based on the information about the reason for treatment, which was summarized into the body part treated as described above. The selection of patients with ankle sprain was based on the medical diagnoses, coded with ICD-9 in FOTO and Maccabi, and with ICPC in LiPZ, both using the same inclusion criteria.
Statistical analyses
Descriptive statistics were calculated for the patient demographic and health characteristics and treatment processes characteristics. In the FOTO database there were over 25% missing cases for the profession of referring physicians variable. Therefore, for this variable the FOTO-data were not used. In all other variables and databases less than 25% missing cases were found. Differences in data were tested using χ2-tests for categorical variables and ANOVA for continuous variables. Differences in the number of treatment visits and treatment duration were tested using linear regression techniques controlling for gender, age and episode duration. To answer the questions about the number of visits and use of interventions, only data of patients for whom the treatment episode was closed were used.
For reasons of readability we used country names instead of database names in the results section.