Table 5.

Summary table of original studies describing or evaluating algorithms or decision-models (theoretical or empirical) for selecting interventions for patients with musculoskeletal disorders

Authors (ID)	Year	Study design	Body part	Algorithm/model mentioned	Population	Methods	Outcome/conclusion	Results
Hurd et al. [63]	2008	Cohort study	Knee	Algorithm for managing subacute anterior ligament (ACL) injuries was created using clinical information on: concomitant injury, unresolved impairments, and results of a screening examination	345 highly active adults (216 men, 129 women) with subacute anterior cruciate ligament injury aged 18–65 years presenting to an orthopedic surgeon	Prospective follow-up study. Patients presenting within 7 months of their injury were treated using a decision-making algorithm. Algorithm was used as criteria to guide management and classify individuals as ‘noncopers’ (poor potential) or potential ‘copers’ (good potential) for non-operative care. Patients were followed up for the duration of care (up to 10 PT sessions over 5 weeks)	199 subjects classified as ‘noncopers’ and 146 as potential ‘copers’. 63 of 88 potential ‘copers’ successfully returned to pre-injury activities without surgery, with 25 of these not undergoing ACL reconstruction at follow-up. The algorithm should be considered as an alternative to management based on anterior knee laxity, age, and preinjury activity levels	Positive
Kodama et al. [66]	2013	Review and retrospective study	Wrist	Scoring system for selecting treatment for distal radius fractures. Includes a variety of clinical factors related to the fracture, as well as dominant hand, high occupational or recreational activity, age, and supplemental factors (Table 2 in paper)	164 patients with distal radius fracture who were 50 years or older presenting to a surgeon.	Development of the decision-making guide was described, and then a retrospective study was used to evaluate the guide in patients. Comparison was made on clinical outcomes (DASH questionnaire scores) between patients where recommendations of the guide were followed and not followed	164 patients were divided into 4 groups using the tool: conservative care, relative conservative care, relative surgical care, and surgical care. Clinical outcomes of those that followed the recommendation were better than those not following the recommendation. The present scoring system is an easy-to-use decision-making tool for choosing conservative or surgical treatment for distal radius fractures	Positive
Murphy et al. [76]	2007	Cohort study	Low back	The approach is based on 3 questions: (1) Are the symptoms reflective of a visceral disorder or a serious/potentially life-threatening disease? (2) From where is the patient’s pain arising? (3) What has gone wrong with this person as a whole that would cause the pain experience to develop and persist?	264 patients with moderate to severe low back pain over 18 years old presenting to a private practice physical therapy clinic	Cross-sectional feasibility study. Demographic, diagnostic and baseline outcome measure data were gathered on a cohort of low back pain patients examined by one of three examiners trained in the application of a diagnosis-based clinical decision rule that guided subsequent treatment	The guide can be applied in a private practice setting It appears that patients with low back pain can be distinguished on the basis of this approach, and treatment plans can be formulated utilizing this strategy	Development article
Sonnabend [64]	1994	Case Series	Shoulder	Treatment algorithm was based on presence of fracture location, weeks in a sling, presence of pain and weakness, and arthrogram or ultrasound results	53 patients with primary traumatic anterior shoulder dislocation older than 40 years of age presenting to an orthopedic surgeon	Patients were classified into 3 groups according to an algorithm based on signs and symptoms. This algorithm was used to determine treatment. Clinical outcomes in the different groups were described after the treatment	The algorithm is suggested as an approach to treatment of primary traumatic dislocation	Development article
Spiegl et al. [65]	2013	Retrospective case series	Shoulder	Treatment algorithm for acute osseous Bankart lesions consisting of a conservative strategy for small defect sizes and a surgical approach for medium-sized and large defects	25 patients who sustained acute traumatic osseous Bankart lesions after a first time shoulder dislocation from a ski or snowboard accident without rotator cuff tears	Retrospective case series to describe outcomes. Operative therapy was performed in patients with osseous defects of 5 % or more, otherwise conservative therapy was initiated	Applying the treatment algorithm appears to lead to encouraging mid-term results and a low rate of recurrent instability in active patients	Positive
Stanton et al. [55]	2011	Cross sectional study and test–retest reliability for a subset	Low back	Treatment-Based Classification Algorithm based on clinical examination findings for selecting treatments for patients with low back pain. This algorithm was summarized into a decision-making flowchart	250 patients with acute or sub-acute low back pain recruited from teaching hospitals (Sydney, Australia) and private physical therapy clinics (Australia and United States)	Observational study to determine the prevalence of patients meeting the criteria for each subgroup (i.e. responders to the various treatments in the system). Trained physical therapists performed standardized assessments on all participants. These findings were used to classify participants into subgroups. 31 participants were reassessed to determine inter-rater reliability of the algorithm decision	Reliability of the algorithm is sufficient for clinical use. But 25 % of participants met the criteria for more than 1 subgroup and 25 % did not meet the criteria for any subgroup. This has important implications for validity and potential revisions to the algorithm’s section that guides unclear classification	Development article
Stanton et al. [56]	2013	Cross-sectional secondary analysis from 3 previous studies	Low back	Treatment-Based Classification Algorithm (see above)	529 patients with low back pain treated at private physical therapy clinics in USA, Australia and the Netherlands, and public hospital physical therapy outpatient departments in Australia	To guide improvements in the algorithm, this study aimed to determine whether people with unclear classifications are different from those with clear classifications. Univariate logistic regression was used to determine which participant variables were related to having an unclear classification	People with unclear classifications appeared to be less affected by low back pain (less disability and fewer fear avoidance beliefs), despite typically having a longer duration of low back pain. Recommendations to the algorithm are suggested, this study provides no evidence that any changes will result in better outcomes	Unclear
Strong et al. [57]	1995	Cohort study	Low back	The Integrated Psychosocial Assessment Model (IPAM), a multidimensional assessment for use with patients with chronic low back pain	70 consecutive patients with chronic low back pain presenting at the Auckland Regional Pain Service or private practice pain facility in Auckland, New Zealand	Subjects were assessed on pain intensity, disability, coping strategies, depression and illness behavior. Cluster analysis was used to identify homogenous groups of patients. Clusters were named “In Control”, “Depressed and Disabled”, and “High Denial”. The clustered obtained by this study were used to propose management	The IPAM model may be valuable for identifying low back pain subgroups. Treatments corresponding to each subgroup were proposed	Development article
Wang et al. [67]	2003	Cohort study	Neck	Clinical reasoning algorithm for treating patients with neck pain. This algorithm was developed before the study by one of the authors. The algorithm consists of 4 categories: (1) radicular arm pain or neck pain; (2) referred arm pain or neck pain; (3) cervicogenic headaches; and (4) neck pain only. There are subcategories formed by different clinical patterns that are used to guide treatment	57 adults referred from general practitioners for physical therapy treatment of neck pain. All patients had current neck pain with or without radiating pain and no other serious pathology	A quasi-experimental, nonequivalent, pretest-post-test control group design was used to investigate the effects of algorithm-based clinical decision making. Outcomes in a treatment group of 30 patients with neck pain treated based on the algorithm were compared to a control group of convenience formed of 27 subjects who also had neck pain but did not receive treatment for various reasons	After ~4 weeks of physical therapy intervention, patients in the treatment group demonstrated statistically significant increases of cervical range of motion, decreased pain, increases of physical performance measures, and decreases in level of disability. The control group showed no differences in all five outcome variables. Authors conclude that organized and specific physical therapy program was effective in improving the status of patients with neck pain, and the algorithm can help clinicians classify patients with cervical pain into clinical patterns	Positive
Widerstrom et al. [58]	2007	Multiple case pretest–posttest study	Back	Clinical ‘pain modulating’ treatment classification for patients with low back pain that was formed empirically. It is considered for patients with moderate to high irritability and high pain and/or disability scores, and where judgments on spinal mobility were inconclusive and no segmental level could be determined	16 consecutive adults patients with low back pain, regardless of duration, with or without radiating pain to the lower extremities. Patients were from the waiting list of a primary care physiotherapy clinic in Sweden. All patients but one had chronic low back pain (>3 months)	The first part of the paper was descriptive, resulting in an individualized clinical decision-making algorithm As an illustration of the utility of the presented algorithm, a multiple subject case study was then conducted, using a pretest–posttest design. The 16 patients were classified based on the algorithm, and treated based on the algorithm, then evaluated at discharge from physiotherapy	Two patients were excluded from the study (1 pregnancy and 1 with progressive symptoms). All but 1 of the remaining 14 patients showed improvements in pain intensity scores. The authors interpret study findings to suggest that the presented model may be used when clinical decisions on selecting interventions for patients with chronic low back pain are made	Positive
Fitzgerald et al. [62]	2000	Cohort Study	Knee	Decision-making scheme for returning patients to high-level activity with non-operative treatment after anterior cruciate ligament rupture. The screening exam consists of four 1- legged hop tests, the incidence of knee giving-way, a self-report functional survey, and a self-report global knee function rating	93 consecutive patients with acute unilateral anterior cruciate ligament rupture	Patients were classified as either candidates (n = 39, 42 %) or non-candidates (n = 54, 58 %) for non-operative management based on the decision-making scheme. Patients were returned to full activity an average of 4 weeks after the screening examination. Successful treatment was defined as the ability to return to preinjury levels of activity without experiencing an episode of giving-way at the knee. Failure was defined as either having at least one episode of givingway at the knee or a reduction in functional status	Of the 39 rehabilitation candidates, 28 chose non-operative management and returned to preinjury activity levels, 22 of whom (79 %) returned to preinjury activity levels without further episodes of instability or a reduction in functional status. The decision-making scheme described in this study shows promise in identifying patients who can safely postpone surgical reconstruction and temporarily return to physically demanding activities	Positive
Rundell et al. [54]	2009	Case series	Back pain	Management of acute and chronic low back pain using the World Health Organization’s International Classification of Functioning. This model provides a method that considers biological, individual, and social contributions that can be used to classify patients	Two patients, 1 with acute and 1 with chronic pain were treated pragmatically using models of clinical reasoning	Manual therapy, exercise, and education interventions were directed toward relevant body structure and function impairments, activity limitations, and contextual factors based on their hypothesized contribution to functioning and disability. Patients were evaluated after a period of 3 and 10 weeks of intervention, respectively	Both patients demonstrated clinically important improvements in pain, disability, and psychosocial factors after intervention. The WHO-ICF model appears to provide an effective framework for physical therapists to better identify each person’s experience with his or her disablements and assists in prioritizing treatment selection	Positive
Shaw et al. [60]	2007	Cohort study	Back pain	A model is developed for discriminating patients with acute back pain into subgroups depending on whether disability is related to pain beliefs, emotional distress, or workplace concerns	528 patients with work-related back pain seeking treatment for acute back pain at one of 8 community-based occupational health clinics in the New England region of the USA	Patients with back pain completed a 16-item questionnaire of potential disability risk factors before their initial medical evaluation. Outcomes of pain, functional limitation, and work disability were assessed 1 and 3 months later	A K-Means cluster analysis of 5 disability risk factors (pain, depressed mood, fear avoidant beliefs, work inflexibility, and poor expectations for recovery) resulted in 4 sub-groups: low risk (n = 182); emotional distress (n = 103); severe pain/fear avoidant (n = 102); and concerns about job accommodation (n = 141). Pain and disability outcomes at follow-up were superior in the low-risk group and poorest in the severe pain/fear avoidant group	Development article
Steenstra et al. [61]	2010	Secondary analysis of previous cohort study data	Back pain	Evaluation of the Risk Factor-Based Intervention Strategy Model proposed previously by Shaw et al. The model was developed based on a literature review and classifies patients into 1 of 4 groups that require different forms of intervention	442 workers with a new, accepted or pending, work related injury lost-time claim for low back pain who were absent from work for at least 5 days within the first 14 calendar days post-injury, and were at least 15 years of age	Claimants (n = 259) who had already returned to work, were categorized as low risk. A latent class analysis was performed on 183 workers absent from work. Groups were classified based on: pain, disability, fear avoidance beliefs, physical demands, people-oriented culture and disability management practice at the workplace, and depressive symptoms	Three classes were identified; (1) workers with ‘workplace issues’, (2) workers with a ‘no workplace issues, but back pain’, and (3) workers having ‘multiple issues’ (the most negative values on every scale, notably depressive symptoms). This study confirms an earlier model theorizing that subgroups of patients can be identified who might benefit from different interventions	Positive but exploratory
Reme et al. [59]	2012	Cohort study	Back pain	Development of a sub-classification of workers with acute back pain. Patterns of early disability risk factors from this study suggest patients have differential needs with respect to overcoming emotional distress, resuming normal activity, and obtaining workplace support	496 workers seeking treatment for work-related, acute back pain at private occupational medicine clinics in the states of Massachusetts, Rhode Island, or Texas, USA	Workers completed self-report measures comprising 11 possible risk factors for chronicity of pain and disability. Outcomes of pain, function, and return-to-work were assessed at 3-month follow-up. A K-means cluster analysis was used to derive patient subgroups based on risk factor patterns, and then these subgroups were compared with respect to 3-month outcomes	A 4-cluster solution met criteria for cluster separation and interpretability, and the four clusters were labeled: minimal risk (29 %), workplace concerns (26 %); activity limitations (27 %); and emotional distress (19 %). Classifying patients in this manner may improve the cost–benefit of early intervention strategies to prevent long-term sickness absence and disability due to back pain	Development article