Abstract
Background Medicaid insures an increasing proportion of adults in the United States. Physical therapy use for low back pain (LBP) in this population has not been described.
Objective The study objectives were: (1) to examine physical therapy use by Medicaid enrollees with new LBP consultations and (2) to evaluate associations with future health care use and LBP-related costs.
Design The study was designed as a retrospective evaluation of claims data.
Methods A total of 2,289 patients with new LBP consultations were identified during 2012 (mean age=39.3 years [SD=11.9]; 68.2% women). The settings in which the patients entered care and comorbid conditions were identified. Data obtained at 1 year after entry were examined, and physical therapy use was categorized with regard to entry setting, early use (within 14 days of entry), or delayed use (>14 days after entry). The 1-year follow-up period was evaluated for use outcomes (imaging, injection, surgery, and emergency department visit) and LBP-related costs. Variables associated with physical therapy use and cost outcomes were evaluated with multivariate models.
Results Physical therapy was used by 457 patients (20.0%); 75 (3.3%) entered care in physical therapy, 89 (3.9%) received early physical therapy, and 298 (13.0%) received delayed physical therapy. Physical therapy was more common with chronic pain or obesity comorbidities and less likely with substance use disorders. Entering care in the emergency department decreased the likelihood of physical therapy. Entering care in physical medicine increased the likelihood. Relative to primary care entry, physical therapy entry was associated with lower 1-year costs.
Limitations A single state was studied. No patient-reported outcomes were included.
Conclusions Physical therapy was used often by Medicaid enrollees with LBP. High rates of comorbidities were evident and associated with physical therapy use. Although few patients entered care in physical therapy, this pattern may be useful for managing costs.
Since its beginning in 1965, Medicaid has grown to become the largest single source of health care coverage in the United States, providing insurance to more than 66 million Americans. Medicaid has experienced continued enrollment growth over the past decade.1 In 2012, total spending on Medicaid was $429 billion.2 Although spending and enrollment vary widely by state and acceptance of the Medicaid expansion provisions in the Patient Protection and Affordable Care Act is incomplete, current projections estimate that 12 million more adults will be covered by 2016.3 It seems clear that the growth of Medicaid will continue as a strategy to reduce the health care insurance coverage gap.
Early research examining Medicaid expansion has been encouraging, demonstrating that access to care and health outcomes improve when people who have been uninsured enroll.4 The continued growth of Medicaid will present substantial challenges for providers and health care systems. Medicaid enrollees are more likely to have multiple and complex medical problems, and people newly insured by Medicaid may be more likely to have undiagnosed or poorly controlled chronic health conditions.5,6 Patterns of health care use differ among recipients of Medicaid, with a reduced likelihood that an individual will have a usual source for primary care and a greater reliance on emergency department (ED) or other safety net settings.7 These concerns may be exacerbated by an influx of people who were previously uninsured.
Musculoskeletal pain complaints, in particular, low back pain (LBP), are among the most prevalent and costly conditions for which recipients of Medicaid seek health care.8 Back pain diagnoses are more common for recipients of Medicaid than for people who are privately insured.8,9 Data from several states have indicated that, compared with people who have private insurance, recipients of Medicaid who have LBP have an increased likelihood of an ED visit and are more likely to be prescribed opioids for pain management.8,10,11
The role of physical therapy in the treatment of Medicaid enrollees who have LBP has not been adequately examined. Traditionally, physical therapy has been an optional benefit in Medicaid, with variable coverage across states and a primary focus on children and adults with disabilities. Medicaid expansion in the Patient Protection and Affordable Care Act, however, mandates coverage for adults who have a low income but do not have children and considers rehabilitation service as an Essential Health Benefit required for coverage; this requirement may expand access to physical therapy to millions of adults who have a low income and were previously uninsured.12 The inclusion of physical therapy as an Essential Health Benefit may result in increased volumes of patients for physical therapists who serve Medicaid enrollees in outpatient settings. Thus, more information on the role of physical therapy is needed.
The objectives of this study were: (1) to examine the use of outpatient physical therapy for LBP by Medicaid enrollees in the state of Utah and (2) to evaluate associations between physical therapy and future health care use and LBP-related costs. Specifically, our 3 aims were: (1) to describe the use of physical therapy and other services for adult Medicaid beneficiaries with new LBP consultations; (2) to evaluate the associations of characteristics of new LBP consultations, including patient factors and provider settings, with the use and timing of physical therapy; and (3) to examine the associations of characteristics of new LBP consultations with the subsequent use of high-cost procedures (advanced imaging, surgery, injection, and ED visit) and total LBP-related health care costs over a 1-year follow-up period.
Method
Data Source
The study sample was identified from claims data of enrollees in the University of Utah Health Plans (UUHP) Healthy U Medicaid Managed Care Plan. The UUHP is a nonprofit health insurance provider and integrated subsidiary of University of Utah Health Care. The Healthy U program covered approximately 43,000 Medicaid beneficiaries in the Salt Lake City, Utah, region. The payment system transitioned to a fully at-risk capitation model beginning January 1, 2013. We matched UUHP member numbers with University of Utah Health Care patient electronic medical record numbers to check for any LBP-related health care use not represented in claims data or other missing patient characteristics.
Sample
We identified patients with new consultations and a billed claim associated with a provider visit and an LBP-related International Statistical Classification of Diseases, 9th Revision (ICD-9), code (Appendix 1) as a primary (first listed) or secondary diagnosis between January 1, 2012, and December 31, 2012. A new consultation was defined as no LBP-related claims in the preceding 90 days. Additional eligibility criteria were an age of 17 to 60 years on the entry visit date and continuous enrollment in Healthy U Medicaid for 90 days preceding and 1 year following the entry visit.
We excluded patients who were pregnant at the entry visit or who had an ICD-9 code indicating a possible nonmusculoskeletal LBP etiology (urinary tract infection [599.0], kidney stones [592.x], or gallbladder stones [574.x]). We excluded patients with a “red flag” condition that may have required urgent management (malignant neoplasm [140.x–209.x], fracture or stress fracture of the spine or pelvis [805.x–809.x, 820.x–821.x, 733.13–733.15, or 733.96–733.98], osteomyelitis [730.x], or cauda equina syndrome [344.6x]). We excluded patients with significant mobility limitations or health conditions that affected their ability to access care and that were identified 90 days before the entry visit or during the 1-year follow-up, including hemiplegia, paraplegia, quadriplegia, wheelchair dependence (344.0x–344.1x, 344.8x–344.9x, or 438.2x–438.5x), or end-stage renal disease (403.01, 403.91, 585.5x–585.6x, or V45.11).
Independent Variables
On the basis of the procedure code and the provider associated with the entry visit, we categorized the entry visit setting as primary care (family medicine, internal medicine, urgent care, or obstetrics/gynecology), physical medicine and rehabilitation, ED, physical therapy, or other setting. We evaluated a 1-year period from the entry visit date to record whether any claims for physical therapy with an LBP-related ICD-9 code were received. We categorized the use of physical therapy on the basis of its position in the patient's care process as follows: physical therapy was the setting for the entry visit (physical therapy entry), physical therapy was not the entry setting but was received within 14 days of entry (physical therapy early), or physical therapy was not the entry setting and was received more than 14 days after entry (physical therapy delayed).
Demographic and Comorbidity Variables
We recorded patient demographic data from UUHP and electronic medical record data, including age at entry visit and sex. Comorbid conditions were recorded from ICD-9 codes for all claims recorded in the year after the entry visit. Specific comorbidities that may influence LBP-related prognosis or health care use were recorded; these included a history of lumbar surgery (V45.4 or 722.83), mental health conditions (296.x–298.x, 300.x, 301.x, 308.x, 309.x, or 311.x), chronic pain (338.2x or 338.4x), substance use disorders (291.x, 303.x–304.x, 305.0, 305.2x–305.9x, or 648.3), smoking status (305.1, V15.82, or 649.0x), and obesity (278.x). We used all comorbidities to compute the Charlson Comorbidity Index in accordance with previously described procedures.13 We used pharmacy claims data to record prescription medication use associated with the entry visit. We examined claims for a 14-day period after the entry visit and recorded prescription medications in the following categories on the basis of therapeutic class codes: nonsteroidal anti-inflammatory medications (S2B), opioid pain medications (H3A, H3M, H3N, H3T, H3U, or H3X), skeletal muscle relaxants (H6H), or oral steroids (P5A).
Outcomes
Outcomes for each patient were evaluated for a 1-year period after the entry visit. We recorded the following health care services as binary outcomes on the basis of Current Procedural Terminology codes in claims data (Appendix 2): radiograph of the lumbopelvic region, advanced imaging (magnetic resonance imaging [MRI] or computed tomography [CT] scan of the lumbopelvic region), surgical procedure (diskectomy, laminectomy, fusion, or rhizotomy of the lumbosacral region), epidural injection of the lumbar spine or sacroiliac joint, and ED visit related to LBP after the entry visit. Costs were recorded from reimbursed costs for all claims associated with an LBP-related ICD-9 code during the year after the entry visit and were summed to compute total LBP-related health care costs.
Data Analyses
Descriptive statistics, including means with standard deviations and frequencies with percentages, were used to describe the sample and physical therapy use (aim 1). We evaluated characteristics of the new consultations for associations with physical therapy use in the following comparisons: any physical therapy versus no physical therapy during the 1-year follow-up period, physical therapy entry versus other entry visit settings, and physical therapy early versus physical therapy delayed (aim 2). Multivariate logistic regression models were used to identify patient demographic and comorbidity variables associated with use. For aims 1 and 2, entry visit setting was included in the model with primary care as the reference category, and the other entry setting was excluded because of the diversity of providers represented in this group. For aim 3, associations of the same set of variables with binary health care use outcomes were examined with multivariate logistic regression models. For comparisons of health care costs between groups, generalized linear regressions with log link function and gamma distribution were considered because of the highly positively skewed distribution of cost data. We accounted for potential clustering effects among patients seeing the same entry provider generating more robust standard errors in the regression models.
Results
A total of 2,289 patients met all eligibility criteria (Figure). Primary care was the most common entry setting (n=1,621; 70.8%); this setting was followed by the ED (n=404; 17.6%), physical medicine and rehabilitation (n=89; 3.9%), and physical therapy (n=75; 3.3%). Other entry settings (eg, orthopedic surgery or neurosurgery, chiropractic, pain medicine) were noted for 100 patients (4.4%). Patient characteristics are outlined in Table 1. Low back pain was the primary diagnosis for 64.6% of the patients. The median number of diagnoses at entry was 3 (range=1–14). Many patients had 2 or more comorbidities (41.8%), on the basis of Charlson Comorbidity Index scores, and 10.0% had 5 or more comorbid conditions. Rates of the comorbidities smoking (24.7%), mental health conditions (61.9%), substance use disorders (15.1%), and chronic pain (32.6%) in the patients were high relative to local averages.
Reasons for exclusion from analyses. LBP=low back pain.
Descriptive Characteristics of Study Samplea
Use of Physical Therapy
Overall, 457 patients (20.0%) received physical therapy at some point during the year after entry. The mean number of physical therapy visits received across the 1-year study period was 5.2 (SD=6.1). Seventy-five patients (16.4% of all patients receiving physical therapy) entered care in physical therapy, 89 (19.5%) received early physical therapy, and 298 (65.1%) received delayed physical therapy (median time to first physical therapy visit=89 days; interquartile range=36–183.5).
Variables Associated With Physical Therapy
The multivariate model for examining variables associated with any physical therapy use (Tab. 2), regardless of timing, revealed that physical therapy was less likely for patients who had a substance use disorder or those who received opioids at the entry visit. Physical therapy was more likely for patients with comorbid chronic pain or obesity. Relative to primary care entry, entry in the ED decreased the likelihood of receiving physical therapy during the year after entry. Entering care in physical therapy was less likely for patients receiving any prescription medication within 14 days but was more likely for women. Variables associated with early versus delayed physical therapy were comorbid chronic pain and entry with a physical medicine provider, relative to entry with a primary care provider (Tab. 2).
Factors Associated With Physical Therapy Usea
Variables Associated With Health Care Use and Cost Outcomes
During the 1-year follow-up period, 710 patients (31.0%) underwent lumbopelvic radiography, 308 (13.5%) underwent advanced imaging, 132 (5.8%) had injections, 30 (1.3%) had surgery, and 348 (15.2%) had an ED visit for LBP beyond the index visit. Variables associated with each use outcome are shown in Table 3. Older age was associated with injections and surgery, whereas younger age was associated with ED visits. With respect to comorbidities, chronic pain was associated with all use outcomes, whereas smoking, obesity, mental health conditions, prescription oral steroids, and prior lumbar surgery each were associated with at least 2 use outcomes. The entry visit setting was associated with use outcomes. Compared with entry in primary care, entry in physical therapy decreased the likelihood of undergoing radiography. No patient entering care in physical therapy had surgery during the 1-year follow-up period. Relative to entry with a primary care provider, entry with a physical medicine physician increased the likelihood of injections or advanced imaging; entry in the ED was associated with radiography, advanced imaging, and additional ED visits (Tab. 3).
Factors Associated With Health Care Use and Cost Outcomesa
Mean total LBP-related costs for all patients during the 1-year follow-up period were $626 (95% confidence interval [CI]=$560, $692). Variables associated with total LBP-related health care costs over the 1-year follow-up period are shown in Table 4. Higher costs were associated with the comorbidities older age, smoking, chronic pain, obesity, mental health conditions, and prior surgery; with prescription oral steroids; and with entering care in the ED; entering care in physical therapy was associated with lower costs. Mean total LBP-related costs were $900 (95% CI=$671.89, $1,129) for patients entering care in the ED, $335 (95% CI=$241, $429) for those entering care in physical therapy, $533 for those with primary care entry (95% CI=$470, $598), and $770 (95% CI=$558, $983) for those with physical medicine entry.
Variables Associated With Total Low Back Pain–Related Health Care Costs Over 1-Year Follow-up Perioda
Discussion
In this study, we examined physical therapy use by Medicaid enrollees with a new LBP consultation and examined health care costs and use in the following year. As anticipated, the sample had high rates of comorbid health conditions, with proportions of smoking, obesity, chronic pain, and mental health conditions much higher than regional averages.14,15 Health care use patterns also differed from those observed for adults with LBP and private insurance, most notably, increased use of the ED.14 The setting in which the patient entered health care was associated with physical therapy use and health care use and cost outcomes. Although only a small number of patients entered care in physical therapy, this pattern was associated with lower 1-year health care costs.
Physical therapy use patterns in the study sample were similar to those in studies involving different payer mixes. In our sample, physical therapy use across the 1-year time frame was 20%, whereas annual rates of 13% to 34% have been reported for Medicare and private-payer samples.16–18 In the present study, 164 patients (7.1%) received physical therapy within 14 days of beginning care, a rate higher than the 3.4% and 6.6% rates reported within similar time frames (14–28 days) for private payers and Medicare, respectively.15,18 A small number of patients (3.3% of the overall sample) entered care in physical therapy. Utah permits direct access to physical therapy, and there are no restrictions on this pattern within state Medicaid policy. This small group tended to be younger, healthier, and more likely to be women than the overall sample, but the factors motivating the decision to enter health care for LBP with a physical therapist cannot be determined. It is possible that these patients had prior experience with physical therapy, a desire to manage their LBP with a more active, exercise-focused approach, or both. Considering the relevance of the choice of entry setting to outcomes and costs in the study sample, further research exploring reasons for patients' choices could lead to strategies to promote more effective and less costly options, such as physical therapy, instead of relying on the ED for entry.
A primary challenge in providing care to Medicaid recipients is the high rate of complex, comorbid health conditions.6 In the study sample, 41% had 2 or more comorbid health conditions, 33% had chronic pain diagnoses, and 62% had at least 1 mental health comorbidity, along with rates of smoking, substance abuse, and obesity that were higher than regional norms. With the exception of substance abuse, these comorbidities were prevalent among patients receiving physical therapy. These factors, particularly mental health conditions and the psychological factors that often accompany chronic pain, can negatively influence outcomes for LBP treatment and increase costs.19–21 The need for physical therapists to provide effective treatment for patients with co-occurring musculoskeletal pain and mental health conditions has been discussed.22–24 Evidence to clarify the most effective practices for patients with complex medical conditions is critical if physical therapists are to play a more prominent role in the treatment of patients who have LBP and are insured by Medicaid.
Actual and anticipated growth of Medicaid has increased attention on managing costs while providing quality care. Identifying factors placing patients at risk for high costs is an important precursor to successful cost management strategies.25 Back pain is a very costly condition among Medicaid recipients,8 and understanding the factors related to incurring high costs therefore may be helpful. Not surprisingly, we found mental health disorders, chronic pain, smoking, and obesity to be important factors associated with LBP-related costs and the use of low-value procedures, such as imaging or ED use, for LBP. These factors are known to drive overall health care costs among Medicaid recipients.25,26 Multidisciplinary programs that concurrently manage mental health or other risk factors (eg, smoking, obesity) along with conditions such as LBP have been advocated as being particularly necessary for Medicaid enrollees, given the high rates of co-occurring chronic health conditions,27 but are challenging to implement.
We additionally found that patients' choice of setting in which to enter health care for LBP was associated with future health care use and LBP-related costs. Relative to entering care with a primary care provider, entering care with a physical medicine physician was more likely to involve imaging or injections; entering care in the ED was associated with higher LBP-related costs in the following year, and patients entering care in the ED were more likely to receive imaging or return to the ED. Conversely, patients entering care in physical therapy had lower LBP-related costs in the following year.
Strategies that influence care choices made by patients—specifically, increasing the numbers choosing to enter care with a physical therapist and avoid the ED—may help to contain costs among Medicaid enrollees with LBP. Use of the ED for nonemergency conditions, such as LBP, is a common concern for Medicaid programs.28 Our results reflect this concern, as patients who began care in the ED and not with a provider office visit had higher health care costs for LBP over the following year. In the present study, 17.6% of Medicaid enrollees with LBP initially sought care in the ED, and 15.2% went within a year.
In our prior work with people who had LBP and private insurance in the same region, about 2% visited the ED for LBP across a 1-year period.14 Our results indicated that beginning care in physical therapy was a preferred option with respect to lower LBP-related health care costs. Patients entering care in physical therapy rather than the ED had LBP-related health care costs that were approximately $550 lower over the following year. As noted previously, our sample included a small number of patients beginning care in physical therapy; however, developing strategies to increase the number of Medicaid participants choosing this option for LBP has the potential to generate large cost savings.
Changing care-seeking patterns requires consideration of both system- and patient-level factors. At the system level, Medicaid enrollees more frequently move between insurance coverages or have periods of noncoverage that can break the continuity of care and influence care seeking.29 Low reimbursement rates for providers and administrative burdens can create access difficulties.30 Patient-level considerations may include work schedules, access to transportation, difficulty managing comorbid psychological conditions, or patterns of learned behavior.31 Improving LBP care patterns for Medicaid recipients requires strategies addressing both system- and patient-level factors that drive ineffective and costly patterns. Such strategies may include shifting hours of operation, educating patients and providers, and increasing the focus within physical therapy on providing self-management strategies for recurrent episodes and exacerbations. Knowledge of barriers and facilitators is required from all stakeholders (policy makers, health care systems, state Medicaid programs, providers, and patients) for the success of strategies to promote more effective, less costly care patterns.
The results of the present study should be considered in light of several limitations. First, the use of claims data precludes the collection of potentially important variables, such as pain severity and duration of symptoms. It is possible that patients who self-selected to attend or enter care in physical therapy had less complicated LBP, possibly confounding the analyses performed. We relied on ICD-9 codes to identify LBP episodes and comorbidities. The accuracy and completeness of ICD-9 coding may be questionable. We used a 90-day period of no LBP claims to define a new consultation. It is likely that some patients were dealing with chronic, recurrent LBP episodes. Finally, we did not have access to patient-centered outcomes, such as pain or function.
Although there is wide variability among states in eligibility criteria for Medicaid, coverage of physical therapy, reimbursement policies, and patient cost sharing,32,33 the projected growth of Medicaid—particularly for adults who are not disabled—makes it likely that physical therapists will see an increased representation of Medicaid in their payer mix.17 This growth will present a variety of opportunities and challenges for physical therapy. The results of the present study examining physical therapy use among Medicaid recipients in one state revealed that physical therapy was used for LBP at rates similar to those for private insurance. High comorbidity levels were evident among patients and were associated with physical therapy use. Few patients entered care in physical therapy, but this pattern was associated with lower 1-year LBP-related health care costs.
Appendix 1.
ICD-9 Codes Used to Identify Low Back Paina
a ICD-9=International Statistical Classification of Diseases, 9th Revision; NEC=not elsewhere classified; NOS=not otherwise specified.
Appendix 2.
CPT-4 Codes Used to Identify Health Care Use Outcomesa
a CPT-4=Current Procedural Terminology, 4th Edition; MRI=magnetic resonance imaging; ICD-9=International Statistical Classification of Diseases, 9th Revision.
Footnotes
Dr Fritz, Dr Kim, and Dr Thackeray provided concept/idea/research design, writing, and data analysis. Dr Thackeray and Ms Dorius provided data collection. Dr Fritz provided project management. Ms Dorius provided institutional liaisons. Dr Kim and Dr Thackeray provided consultation (including review of manuscript before submission).
The study was approved with exempt status by the University of Utah Institutional Review Board.
- Received January 26, 2015.
- Accepted August 18, 2015.
- © 2015 American Physical Therapy Association