Development and Application of Clinical Prediction Rules to Improve Decision Making in Physical Therapist Practice

Clinical prediction rules (CPRs) are tools designed to improve decision making in clinical practice by assisting practitioners in making a particular diagnosis, establishing a prognosis, or matching patients to optimal interventions based on a parsimonious subset of predictor variables from the history and physical examination.1,2 Clinical prediction rules have been developed to improve decision making for many conditions in medical practice, including the diagnosis of proximal deep vein thrombosis (DVT),3 strep throat,4 coronary artery disease,5 and pulmonary embolism.6 Clinical prediction rules also have been developed to assist in establishing a prognosis such as determining when to discontinue resuscitative efforts after cardiac arrest in the hospital,7 determining the likelihood of death within 4 years for people with coronary artery disease,7 identifying children who are at risk for developing urinary tract infections,8 and identifying the characteristics of patients who are likely to develop postoperative nausea and vomiting after anesthesia.9 Clinical prediction rules have recently been developed that can improve decision making in physical therapist practice. Examples include prediction rules to improve the accuracy of diagnosing ankle fractures (ie, “the Ottawa Ankle Rules”)10 and knee fractures (ie, “the Ottawa Knee Rules”)11 in people with acute injuries and to determine when to order radiographs in patients with neck trauma.12 Other prediction rules have been developed to diagnose patients with cervical radiculopathy13 and carpal tunnel syndrome.14 A CPR also has been developed to establish the prognosis of patients with neck pain following a rear-end motor vehicle accident.15 With increasing attention focused on the rising costs of health care, CPRs provide practitioners with powerful diagnostic information from the history and physical examination that may serve as an accurate decision-making surrogate for more expensive diagnostic tests. For example, the Ottawa Ankle Rules identify only those patients in which the probability of having a fracture is sufficiently large to warrant radiographic imaging, thus reducing costs and avoiding exposing patients to unnecessary radiation.16 Similarly, if the CPRs used to diagnose patients with cervical radiculopathy13 and carpal tunnel syndrome14 are eventually validated, the demand for electrodiagnostic testing may be reduced, potentially saving costs and avoiding the discomfort and anxiety associated with these procedures. In addition to their diagnostic utility, CPRs pertinent to physical therapist practice have recently been developed to assist with subgrouping patients into specific classifications that are useful in guiding management strategies. For example, CPRs have been developed to help practitioners match patients to optimal treatment approaches such as spinal manipulation17,18 and a lumbar stabilization exercise program.19 An advantage of CPRs is that they use the diagnostic properties of sensitivity, specificity, and positive and negative likelihood ratios (LR); thus, their interpretation can be readily applied to individual patients.1 Although helpful for guiding the early stages of treatment and assigning patients to a particular classification, they are not always useful for prescribing the exact treatment techniques to be used within the context of the patient’s assigned classification. Because CPRs are designed to improve decision making, it is important that they be developed and validated according to rigorous methodological standards. McGinn et al1 have suggested a 3-step process for developing and testing a CPR prior to widespread implementation of the rule in clinical practice. The purpose of this update is to describe the different steps involved in developing and validating CPRs and illustrate how CPRs can be used to improve decision making in physical therapist practice. Clinical prediction rules have the potential to improve outcomes, increase patient satisfaction, and decrease costs of care in physical therapist practice. The initial step in the development of a CPR involves creation of the rule (Fig. 1). Researchers and practitioners may initially brainstorm to develop a list of all possible factors that they believe have some predictive value for identifying the condition of interest. Ultimately, a reasonable list of predictors are selected for consideration based on clinical experience and previous research, which demonstrates that the factor or set of factors has some diagnostic or prognostic accuracy. Although it may be ideal to include every possible factor from the clinical examination to ensure that no possible predictor variables are overlooked, the researcher must weigh the benefits of including a complete set of potential predictor variables against the increase in sample size required for each additional variable under consideration. Some authors20,21 have recommended that 10 to 15 subjects should be enrolled into the study to identify one predictor variable. The sample size also must be judged in the context of the risks and benefits of decision making based on the rule and the prevalence of a particular phenomenon. For example, there may be significant consequences associated with the failure to identify a clinically relevant cervical spine injury in a patient who has sustained neck trauma or with the failure to identify the presence of an ankle fracture. These studies, therefore, tend to enroll thousands of patients to achieve sufficiently narrow confidence intervals so that practitioners can be virtually certain that application of the rule will not lead to an error in decision making. These injuries also are relatively rare in light of the total number of traumatic injuries; thus, larger sample sizes are necessary to observe a sufficient number of events on which to base the accuracy calculations. Steps in the development of a clinical prediction rule. Steps in the development of a clinical prediction rule. On the other hand, although failing to identify a patient likely to benefit from a specific treatment approach such as spinal manipulation may result in a less-than-optimal outcome or a delay in improvement, the patient is unlikely to have a serious complication. Furthermore, the pretest probability of achieving a successful outcome with spinal manipulation (ie, the probability associated with a successful outcome before considering the patient’s status on the rule) was 45%, which is considerably higher than the prevalence of a clinically relevant cervical spine injury or ankle fracture among individual patients with an acute injury. This also permits a smaller sample size because fewer cases are necessary to observe a sufficient number of events (ie, a successful outcome) to characterize the accuracy of decision making within an acceptable level of confidence. The development of CPRs, therefore, requires the researcher to consider the prevalence that a particular event will occur and then balance the benefits of achieving ever more narrower confidence intervals against the additional costs associated with recruiting an increasingly large sample size. Once the initial set of possible predictor variables is established, patients are examined to determine the presence or absence of each predictor variable at baseline. To minimize bias, it is essential that the examiner be blinded from knowing whether the patient actually has the condition of interest. For example, in the development of the spinal manipulation CPR, a variety of demographic, historical, and physical examination findings were considered.18 Patients with low back pain first completed several questionnaires consisting of self-report measures of pain and function. Factors such as the mechanism of injury, nature of current symptoms, distal extent of symptoms, and previous episodes of low back pain were considered as possible predictor variables. In addition, psychosocial considerations such as fear-avoidance beliefs and nonorganic signs and symptoms were considered. Physical examination findings that were considered to be potential predictor variables included hip and lumbar spine range of motion, lumbar spine mobility testing, and a variety of traditional landmark symmetry and provocation tests purported to identify dysfunction in the lumbopelvic region. After patients are examined at baseline for the presence of the possible predictor variables, a second examiner who is blinded to the results of the clinical examination should then establish whether the patient actually has the condition of interest according to a standardized and well-accepted reference criterion (ie, “gold standard” or “reference standard”). Reid et al22 suggested that an appropriate reference criterion is one that accurately represents the condition the diagnostic test is attempting to identify. For the spinal manipulation CPR, the purpose was to “diagnose” patients with low back pain who were likely to achieve a dramatic improvement from spinal manipulation after 1 week.18 The reference criterion, therefore, was based on response to a standardized manipulative intervention according to a predetermined clinically relevant cutoff score. In this case, patients who achieved at least 50% improvement on the Oswestry Low Back Pain Disability Questionnaire, a patient’s perceived level of disability, were considered to have achieved a successful outcome. Previous research23–25 has shown that 50% improvement on the Oswestry Low Back Pain Disability Questionnaire distinguishes between patients responding to manipulation versus those simply benefiting from the favorable natural history of low back pain. During development of the Ottawa Ankle Rules10 and Ottawa Knee Rules,11 radiographs were the reference criterion to determine whether an ankle or knee fracture was present. Neural conduction using electrodiagnostic testing based on well-established guidelines for the diagnosis of carpal tunnel syndrome and cervical radiculopathy was used as the reference criterion to determine which patients actually had the condition.13,14 The credibility and usefulness of the CPR that is eventually developed hinges upon the selection of an appropriate and clinically meaningful reference criterion. Once the collection of possible predictor variables and blinded determination of the presence of the condition of interest have been the can be Although other techniques may be is a used approach to determine the parsimonious set of predictor variables in a CPR that the accuracy of diagnosing the condition of The of how to are the of this a more of this is in et In the accuracy of CPRs is using diagnostic accuracy such as sensitivity, specificity, and positive and negative of these are a is With to CPRs, is the of patients with the condition or outcome of interest who are positive on the CPR (ie, positive the of a test to identify the condition when present. is the of patients who not have the condition or outcome of interest and are negative on the CPR (ie, negative the of a test to when the condition or outcome of interest is ratios the information from and A positive the in the diagnosis or outcome when the patient the of the CPR (ie, to rule in the a negative the in the diagnosis or outcome when the patient not the (ie, to rule the An accurate CPR, therefore, have a large positive to rule in the diagnosis, or a negative to rule the to et accuracy can be considered when a positive is than or a negative is than is when a positive is than or a negative is than a CPR can be recommended for use in clinical practice, it is necessary to the CPR in a or to ensure that results are in a different of patients or in a different health care (Fig. is possible that some of the predictor variables that in the development may have by This is because the of identifying predictive variables may not consider whether the factors are The a and to be predictive of a to a error in the for also can light on whether the predictors For example, it that patients who are likely to benefit from spinal manipulation may tend to have more acute symptoms, a of symptoms that not distal to the low fear-avoidance and some of in the lumbar there is no these factors will in a different of Clinical for of having a proximal and the deep is by in the of the the and the of the vein in the and with a 10 are or postoperative Clinical for of having a proximal and the deep is by in the of the the and the of the vein in the and with a 10 are or postoperative in a development also may be to a particular of patients or the practitioners who in the this is the case, the predictor variables may not be to other patient or different For example, and have developed a CPR to identify patients of having a proximal 1). A patient’s on the rule is by each judged to be A of to a low probability of having a proximal a of 1 or to a and a that is than or to to a probability that a proximal attempting to the rule included a of with a range of making it whether the rule was accurate among with one of the likely to experience a proximal et this in a study by considering the accuracy of the rule in a of patients with including patients who were from who had sustained a traumatic injury or or who had been with one or more of the spine or The included patients who had sustained who were from or who had a as to whether these patients had a or risk for proximal were with the in the of patients with a range of for the use among with conditions to have proximal confidence of patients in the low probability confidence in the probability and confidence in the probability were to have a proximal The that the confidence intervals in the different probability not also was as to the in this of The results of this study the for and that physical and other health care should use the CPR to with conditions who are at risk for proximal A it is important to is that a different of practitioners may to accurately the CPR or may the tests and measures used to determine the presence of predictor variables in the CPR than in the initial practitioners in how to the examination and treatment in is essential to the that a useful CPR will in a For example, the study for the spinal manipulation involved physical from a variety of different in study and of the manipulation the Ultimately, a CPR is useful only to the extent that it can improve clinically relevant outcomes, increase patient satisfaction, and decrease costs it is into the of clinical practice. The step in the development of a CPR, therefore, involves the of implementation on practice of care, and costs (Fig. 1). are in 1 of individual patients be assigned to care based on the CPR or have that are based on practice. this requires that practitioners back and between decision making based on the CPR and decision making based on care, which may not be clinically in practice An approach is to clinical to the CPR or not for all A is to use a in which are within the practice before and after the Although the is a reasonable the of the findings is with the The Ottawa Ankle Rules provide a for how successful can be et assigned to the Ottawa Ankle Rules or use of decision making. a traditional ankle radiographs were in of patients with only of patients when decision making was based on the Although the using the Ottawa Ankle Rules ankle were associated with an outcome. In a et a in the of ankle radiographs and a in radiographs upon implementation of the Ottawa Ankle Rules with a not in the implementation of the rule with patients who radiographs not have a patients also in the versus and had total medical costs versus The of patients who were with their care was between the versus these results were achieved the of care and were a after the in costs of care, and patient or of care have been with the implementation of the Ottawa Knee should also the of with CPRs in different health care medical versus health versus to determine whether the rule between different health care The application of CPRs in physical therapist practice can have important for decision making and can assist practitioners in making potential treatment is important to that the diagnostic accuracy of interest will on the The of to positive and negative of which may lead to an error in decision making. the therefore, is not the when one may have more consequences than the For example, with the Ottawa Ankle practitioners not to patients with an ankle therefore, it was essential to minimize the number of negative These negative findings be patients for the rule radiographs are when in they have an ankle fracture. the positive in this the of radiographs at the of several patients in which a fracture in the associated with a additional radiographs judged to be (ie, a positive is the potential risks associated with a fracture (ie, a negative In these the of interest is to the sensitivity, so that negative on the CPR virtually the that an ankle fracture therefore, can be in a decision not to order radiographs for patients who are negative on the CPR, because they that they are not exposing their patients to risk of an from a failure to a fracture. A has been used in the development of CPRs to determine when radiographs are for patients who have a serious knee or trauma to the are to the by et for a on the use of to determine in the probability that a patient has the condition or outcome of interest. a by is a useful that can be used to the and thus is for use in a clinical practice. A of how to use this has been The use of CPRs to treatment approaches for individual patients can be by the spinal manipulation and lumbar stabilization For the spinal manipulation CPR, of patients achieved at least a 50% improvement in their Oswestry Low Back Pain Disability Questionnaire of whether they the on the if practitioners were to provide manipulation to patients with low back of patients will have at least a 50% improvement in within 1 The study to identify patients who likely benefit from spinal thus, the of interest was the positive The positive was for patients who at least 4 of the To this result in the probability of achieving a successful outcome from to therefore, practitioners can be increasingly certain that patients who at least 4 of the in the CPR will achieve at least 50% improvement in by the of 1 in the Clinical in the Clinical With the positive was which into a probability of the with which this manipulation can be and in light of the low an at spinal manipulation may be The study in findings to the on a pretest probability of of and a positive of a patient who is positive on the rule and with manipulation had a of achieving a successful outcome by the of 1 A patient’s status on the rule was of in determining the outcome of patients with an exercise the that the rule is a response to spinal A CPR has been developed to identify patients who are likely to benefit from a lumbar stabilization exercise approach of the subjects had at least 50% improvement on the Oswestry Low Back Pain Disability Questionnaire after of a exercise if practitioners were to lumbar stabilization for patients with of patients will achieve at least 50% improvement in by the of the positive was among patients with at least of 4 predictors of thus, the probability of a achieving at least 50% improvement to The also the likelihood that a patient have improvement as patients who than 50% improvement on the Oswestry Low Back Pain Disability Questionnaire, achieved an improvement that the clinically important of with the lumbar stabilization The that if a patient or more of the the likelihood that the patient clinically meaningful improvement with lumbar stabilization was in the Clinical for the of and in the Clinical for the of and Clinical prediction rules also can be used to help practitioners determine when a particular treatment approach may not be For example, the probability of a successful outcome among patients with than of on the spinal manipulation CPR is no than if a were to provide manipulation to patients with low back pain In a the negative for patients who fewer than of the in the CPR was confidence reducing the probability of to only With fewer than the probability of responding to manipulation approaches that practitioners may to consider other treatment approaches with a higher probability of making can be by identifying clinical variables associated with the failure of patients with to improve from spinal A CPR also has been developed to identify patients who are likely to to to a lumbar stabilization exercise approach In this was as than of improvement on the Oswestry Low Back Pain Disability Questionnaire, which has been shown to be the clinically important for this this criterion, of patients with low back pain will to improve at least on the Oswestry Low Back Pain Disability Questionnaire after The likelihood of improvement with a lumbar stabilization to among patients with one or of the variables in the in the carpal tunnel syndrome clinical prediction in the carpal tunnel syndrome clinical prediction in the cervical radiculopathy clinical prediction in the cervical radiculopathy clinical prediction CPRs to improve diagnostic decision making can be by the prediction rules designed to improve the accuracy of diagnosing cervical radiculopathy13 and carpal tunnel In these the was to rule in the diagnosis of these thus, the of interest was the positive The powerful of factors for in the diagnosis of carpal tunnel syndrome is in on a pretest probability of that the patient may have carpal tunnel syndrome and a positive of when at least 4 of findings are the probability of having carpal tunnel syndrome to a positive of when all findings are the probability to An of the patient’s status on the therefore, can help the diagnostic process for determining whether the patient has carpal tunnel decision-making can be applied to the cervical radiculopathy CPR (Fig. a pretest probability of and a positive of for patients with at least of 4 findings the probability of having a cervical radiculopathy is to all 4 findings are a positive of the probability to increasing the level of confidence that the patient has a cervical advantage of CPRs is in addition to decision making based on the patient’s status on the the individual variables that the rule to be For example, the which is by the by the based on and is to be an of carpal was the useful test for carpal tunnel syndrome when the value The test by was the test for establishing a diagnosis of cervical With a negative to a negative test rules the presence of cervical of for Clinical of for Clinical McGinn et have an to assist practitioners in determining whether the CPR is appropriate for use in the decision-making Clinical prediction rules in which the rule has been not validated are as level For example, the carpal tunnel syndrome and cervical radiculopathy prediction rules to a level CPR, the first step in the development of the rule. are necessary before these rules can be recommended for widespread use in clinical practice. A level CPR is one that has only been validated in one narrow and thus should be used with among patients in a practice Clinical prediction rules be recommended for widespread implementation at least one large study in a of patients and practitioners has been in a variety of practice For example, the spinal manipulation CPR study with a variety of experience in different health care thus, the it as level on the This the confidence that the spinal manipulation CPR can be used in a of patients with low back pain to improve decision making and patient The study of the CPR to identify with conditions who have a proximal also is with a level increasing a confidence in the accuracy among patients with A level CPR to the level of and requires at least one study in a different results from an study in practice of care, and costs For example, practitioners can be that decision making based on the Ottawa Ankle Rules will not only be accurate in a variety of health care will also the at which ankle radiographs to be thus reducing health care reasonable to believe that for a level or level CPR should be sufficient to practice their a to be with the is a having a level CPR such as the Ottawa Ankle Rules not that it can be into clinical practice. and no in the use of ankle among who had been in the use of the Ottawa Ankle The for practitioners is to an to CPRs in a clinical are required to the individual factors in the CPR and how to patients with to each criterion, and they must in the context of the decision-making process to the accuracy of their CPRs that have many predictor variables may be for the to and in clinical practice. practitioners are that the CPR is to use and will improve costs or of care, implementation may be to to not to thus, efforts should be to implementation that standardized practice that are with the to the of each practice are Clinical prediction rules have the potential to improve outcomes, increase patient satisfaction, and decrease costs of care in physical therapist practice. can be useful tools to practitioners and to patients their diagnosis or development has to among health care that diagnostic tests such as radiographs or tests provide useful for decision making, information from the patient history is more and not as they also may be useful to increase the of clinical by to study more of

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.