Evaluation of Parity in the Federal Employees Health Benefits (FEHB) Program: Final Report. Impact on Quality of Care

12/31/2004

The PERT studied the effect of parity on quality of care for adults by using established standards of care for treating specific disorders, i.e., depression and substance abuse, to create indicators of quality following the methods developed by several investigators (c.f. Berndt et. al 1997; Lehman and Steinwachs, 1998). The methods described in this section use claims and encounter data to study changes in indicators of quality of MH/SA services attributable to implementing parity for MH/SA coverage within the FEHB Program.

Although efficacious treatments exist for many common MH/SA disorders, the quality of care actually provided is sometimes inadequate. Research indicates, for instance, that more than two-thirds of people with a depressive or anxiety disorder receive no appropriate treatment (i.e., no treatment or inappropriate treatment), even when they have medical insurance (Young et al., 2001).

Key Research Question

Chapter III reported on several FEHB plan efforts to implement quality improvement strategies. This chapter presents the evaluation of the impact of the parity policy on quality over the period of the study. The principal question that guided the evaluation of the impact of quality was: How well do the patterns of care for MH/SA conditions (as evidenced in claims/encounter data) reflect adherence to established treatment guidelines?

For a complete discussion of the archival data collection, see Data Collection under Impact on Access to Care, Service Use, and Cost.

General Analytic Methods

The claims-based approach was used to directly compare quality before and after implementing parity, but was limited to individuals who had a claim for MH/SA treatment. This approach estimated the expected outcomes (effectiveness) of care provided to individuals who were in active MH/SA treatment.

Epidemiologic studies indicate that this population represented only some of the people who needed care. For example, while this population included some individuals who received MH/SA treatment only in primary care, others who received MH/SA care in this setting may have gone undocumented.

Impact on Quality of Care for Depression

Overview and Model

PERT researchers focused on depression as a “tracer” condition since it typically accounts for about 50% of spending for MH/SA care in private insurance and effective treatment for this disorder has been well documented (Wells et al., 1996). The PERT limited the analysis to major depressive disorder (MDD), rather than all depressive disorders because prior research experience suggests that the “false positive” rate (i.e., the proportion of persons diagnosed with a disorder compared to those who meet research criteria for the disorder) is lower for MDD compared to other ICD-9-CM31 depressive disorder categories.

The PERT measured potential changes in quality of depression care based on service users receiving guideline recommended treatments in a given claims year as well as within the acute phase of MDD (i.e., the first four months of an episode). Year-level analyses examined the likelihood of individuals diagnosed as having MDD receiving guideline-recommended psychopharmacologic or psychotherapeutic treatments. Acute-phase, episode-level analyses measured the dose and duration of those treatments for MDD-diagnosed enrollees. For both the yearly and phase-level analyses, separate models were examined for each plan.

Characterizing treatments received in episodes of treatment is important when assessing quality of MDD care. This is because MDD is often characterized by exacerbations and remittances. Acute phase treatment is clinically defined as the duration of treatment needed until a service user has complete resolution of depressive symptoms. Often, this phase is categorized by more intensive follow-up and, if medication is prescribed, changes in dose or type of medication. Typically, in clinical efficacy trials, this phase is expected to last at least three months.

In prior claims analyses, PERT researchers had considered acute-phase usual care treatment to last four months in order to account for possible inefficiencies in usual care settings (e.g., missed appointments). In clinical settings, continuation phase treatment begins once a service user has been stabilized and his or her symptoms have resolved.

From a quality assessment perspective, one would then expect a different intensity of treatment based on whether a service user is in the acute phase of an exacerbation instead of the continuation phase. Examining at the episode-level provides important information about quality that can take into account a minimum appropriate intensity and duration of treatment, allowing for a more nuanced assessment of quality than does a year-level analysis alone.

Specific Methods for Quality of Care for Depression Analysis

Service users diagnosed as having MDD were identified by ICD-9-CM code using an algorithm-based approach previously used by the PERT. This algorithm aimed to balance minimizing the false positives with maximizing the true positive rates of identifying people as having MDD.

Once accepted into the MDD analysis cohort, service users were then classified according to clinical and demographic information contained in enrollment files and claims and encounter data. Key variables used in classifying patients were:

  • demographic characteristics (i.e., age, gender, and relationship to the contract holder),
  • diagnostic information (i.e., presence of co-occurring substance use disorders or other psychiatric), and
  • presence of intensive mental health treatment in that year and episode (i.e., inpatient, partial hospital, or residential treatment).

New acute phase MDD episodes were determined by either:

  • a new ICD-9-CM code for MDD (i.e., 296.2 or 296.3) after a three-month period of no MH/SA treatment, or
  • an inpatient hospitalization for MDD.

Acute phase treatment was determined to “end” when one of the following occurred:

  • the service user dropped out of MDD treatment (i.e., no MH/SA treatment for at least 90 days),
  • an MH/SA inpatient admission occurred,32 or
  • four months of acute phase MDD treatment had been received.

Analytic Strategy

The quality indicators were informed by guidelines published by the American Psychiatric Association and the Agency for Healthcare Research and Quality (Agency for Health Care Policy and Research, 1993; Agency for Health Care Policy and Research, 1999; American Psychiatric Association, 2000). We included person-year and acute phase episode level analyses. In specifying the dose and duration of psychotherapy and antidepressants, the guidelines typically recommend “as clinically indicated” based on a patient’s illness severity and response to treatment. (Appendix C lists the medication for major depressive disorders.) Because this information is not knowable in claims data analyses, we also included measures that have clinical face validity as minimum quality standards for new onset acute phase episodes of MDD.

Specifically, in the person-year analyses, we examined the likelihood of receiving any antidepressants, any psychotherapy, or either treatment in a given year, i.e., pre-parity (years 1999 and 2000) and post-parity (years 2001 and 2002). Receiving either (or both) antidepressants and psychotherapy would be appropriate treatment for MDD, according to the quality guidelines. However, recognizing that the FEHB parity policy may affect these two treatment modalities differently, we also measured changes in receiving psychotherapy separately from changes in receiving antidepressants.

Enrollees were included in the person-year analyses each year that they received at least one MDD diagnosis. In the acute phase episode level analyses, we examined the duration and intensity of MDD treatment follow-up, psychotherapy, and antidepressant medication. Acute phase was defined as the first four months of treatment. In the episode level analyses, we also focused on the acute phase (rather than longer-term follow-up and treatment phases) because that period is associated with the most intensive treatment need. Specifically, we examined:

  1. Duration of mental health follow-up treatment (visits and/or medications),
  2. Intensity of follow-up visits within the first two months and in the second two months of acute phase treatment,
  3. Duration and intensity of psychotherapy, and
  4. Cumulative antidepressant treatment duration (i.e., total number of days of antidepressant use in acute phase that may or may not be continuous).

Each of the seven plans in this section was analyzed separately.33 Only MDD-diagnosed enrollees who were continuously enrolled for all four study years (1999 through 2002) were included in the analysis.

Each of the regression models controlled for age, gender, relationship to the health plan policy contract holder, and the presence of a mental health co-occurring condition34. The person-year models also controlled for co-occurring substance use disorders excluding tobacco use disorders. We were unable to control for substance use disorders in the acute phase episode analyses because of a low detected prevalence of substance use disorders in the acute phase episodes.

We constructed 95% confidence intervals for the odds ratios based on the estimated standard errors. The standard errors were derived from the application of the Generalized Estimating Equation (GEE) approach. In the episode level analyses, few enrollees were treated for MDD acute phase episodes in both the pre- and post-parity implementation periods. Therefore, this analysis is more similar to a cross-sectional rather than a longitudinal design in which enrollees serve as their own controls.

Findings on Quality of Care for Depression

Person-year Analyses

Table IV.L.1 shows the MDD diagnosis identification rates per plan among all continuously enrolled enrollees for each of the years 1999 to 2002. Increases in MDD diagnosis rates over the four years did not exceed 0.5% except for HMO-W1, which increased from 2.3% to 3.2%.

Table IV.L.1. MDD diagnosis identification rates among all continuously enrolled enrollees (actual percentages)

Plan Pre-parity Post-parity
1999 2000 2001 2002
FFS-MA1 2.5 2.6 2.8 2.8
FFS-MA2 2.6 2.8 3.0 3.0
FFS-NE1 1.5 1.7 1.8 1.9
FFS-NE2 1.3 1.4 1.6 1.6
FFS-W 2.4 2.5 2.8 2.9
FFS-S 2.0 2.1 2.2 2.3
HMO-W1 2.3 2.7 3.0 3.2

Table IV.L.2 below describes the person-year sample size per selected FEHB plan. The percent of the sample pre-parity and post-parity are shown. (NB: Percent pre-parity plus percent post-parity add to 100 percent.)

Table IV.L.2. Person-year sample size by plan

Plan Pre-parity Post-parity
Frequency Percent Frequency Percent
FFS-MA1 5,605 48.30 6,000 51.70
FFS-MA2 4,043 47.19 4,525 52.81
FFS-NE1 1,224 46.24 1,423 53.76
FFS-NE2 593 46.33 687 53.67
FFS-W 2,538 46.38 2,934 53.62
FFS-S 2,814 47.82 3,071 52.18
HMO-W1 903 45.08 1,100 54.92

Actual Proportion Receiving Services Consistent with Quality Measures (Antidepressants or Psychotherapy)

Table IV.L.3 describes the actual (i.e., unadjusted) proportion of person-years in which the quality measures were received in each plan. Across plans, there was variability in the proportion receiving services consistent with each of the quality measures. In the post-parity period, the frequencies of receiving these quality measures were similar. The exception was HMO-W1, which experienced a 12.3 percentage point increase in the proportion receiving any psychotherapy. After adjusting for enrollee characteristics, though, this was not a significant change, as we will see in the regression results.

Table IV.L.3. Proportion of MDD diagnosed enrollees who received any psychotherapy or antidepressant

Plan Pre-parity Post-parity
MDD Diagnosed Enrollees Received Any Psychotherapy or Antidepressant MDD Diagnosed Enrollees Received Any Psychotherapy or Antidepressant
FFS-MA1 5,205 92.9% 5,646 94.1%
FFS-MA2 3,656 90.4% 4,180 92.4%
FFS-NE1 1,079 88.2% 1,291 90.7%
FFS-NE2 542 91.4% 638 92.9%
FFS-W 2,261 89.1% 2,696 91.9%
FFS-S 2,498 88.8% 2,824 92.0%
HMO-W1 791 87.6% 998 90.7%

At least 90% of the MDD diagnosed enrollees received either an antidepressant or psychotherapy.

As shown in Table IV.L.4 and Table IV.L.5, in both the pre- and post-parity periods, receiving antidepressants was more prevalent across plans (about 75% of enrollees diagnosed with MDD) than was receiving any psychotherapy (about 50% of diagnosed enrollees). As Table IV.L.4 shows, for all seven plans, a slightly greater proportion of MDD diagnosed enrollees receive an antidepressant in the post-parity period than in the pre-parity period. In six out of seven plans, the same was true for MDD diagnosed enrollees who received any psychotherapy. In FFS-MA1, MDD diagnosed enrollees were slightly less likely to receive any psychotherapy in the post-parity period.

There was more variability across plans in the prevalence of receiving any psychotherapy compared to the other quality measures.

Table IV.L.4. Proportion of MDD diagnosed enrollees who received any antidepressant

Plan Pre-parity Post-parity
MDD Diagnosed Enrollees Received Any Antidepressant MDD Diagnosed Enrollees Received Any Antidepressant
FFS-MA1 4,461 79.6% 4,913 81.9%
FFS-MA2 3,231 79.9% 3,699 81.8%
FFS-NE1 885 72.3% 1,080 75.9%
FFS-NE2 413 69.7% 520 75.7%
FFS-W 1,899 74.8% 2214 75.5%
FFS-S 2,253 80.1% 2,533 82.5%
HMO-W1 685 75.9% 840 76.4%

 

Table IV.L.5. Proportion of MDD diagnosed enrollees who received any psychotherapy

Plan Pre-parity Post-parity
MDD Diagnosed Enrollees Received Any Psychotherapy MDD Diagnosed Enrollees Received Any Psychotherapy
FFS-MA1 3,617 64.5% 3,683 61.4%
FFS-MA2 1,989 49.2% 2,293 50.7%
FFS-NE1 657 53.7% 805 56.6%
FFS-NE2 380 64.1% 453 65.9%
FFS-W 1,373 54.1% 1,729 58.9%
FFS-S 1,134 40.3% 1,359 44.3%
HMO-W1 309 34.2% 511 46.5%

Estimations of Likelihood of Receiving Quality Measures

Tables IV.L.6, IV.L.7, and IV.L.8 describe the adjusted odds of receiving the quality measures in the post-parity (2001 and 2002) period relative to the pre-parity (1999 and 2000) period. The models controlled for co-occurring SA disorders (excluding tobacco disorders), age, gender, relationship to contract holder, and presence of a co-occurring MH condition. With the exception of HMO-W1, all plans experienced some quality improvement associated with the post-parity implementation period.

FFS-MA1 was the only plan to experience a quality decrement: In FFS-MA1, MDD-diagnosed enrollees were less likely to receive any psychotherapy in the post-parity period. However, post-parity they were more likely to receive antidepressant medication and more likely to receive either antidepressant medication or psychotherapy (i.e., at least one of the guideline recommended treatments). Thus, on balance, FFS-MA1 enrollees experienced a greater likelihood of receiving any MDD treatment that conformed to the quality measures after implementation of the parity policy. This change was due to an increase in antidepressant prescribing, while the likelihood of receiving any psychotherapy decreased. It is notable that in the pre-parity period, FFS-MA1 had the highest proportion of persons receiving any psychotherapy of all the plans (approximately 64%). Post-parity, this proportion declined slightly (to approximately 61%) but was still among the highest of the plans. Other plans in which quality improvements in receiving either treatment modality were observed include FFS-W, FFS-S, and FFS-MA2.

Table IV.L.6. Regression results for adjusted odds ratio of receiving any psychotherapy or any antidepressant in the post-parity period relative to the pre-parity period

Plan Odds ratio Confidence interval
FFS-MA1 1.23*** 1.09, 1.39
FFS-MA2 1.26*** 1.11, 1.43
FFS-NE1 1.20 0.95, 1.52
FFS-NE2 1.18 1.85, 1.65
FFS-W 1.26** 1.07, 1.48
FFS-S 1.36**** 1.18, 1.57
HMO-W1 1.07 0.82, 1.38
** p<0.01
*** p<0.001
**** p<0.0001

 

Table IV.L.7. Regression results for adjusted odds ratios of receiving any antidepressant in the post-parity period relative to the pre-parity period

Plan Odds ratio Confidence interval
FFS-MA1 1.14**** 1.07, 1.22
FFS-MA2 1.14** 1.05, 1.23
FFS-NE1 1.21** 1.05, 1.40
FFS-NE2 1.34** 1.11, 1.61
FFS-W 1.06 0.97, 1.17
FFS-S 1.14** 1.03, 1.26
HMO-W1 1.00 0.85, 1.18
** p<0.01
**** p<0.0001

 

Table IV.L.8. Regression results for adjusted odds ratio of receiving any psychotherapy in the post-parity period relative to the pre-parity period

Plan Odds ratio Confidence interval
FFS-MA1 0.87*** 0.81, 0.94
FFS-MA2 1.02 0.94, 1.12
FFS-NE1 0.96 0.82, 1.13
FFS-NE2 0.93 0.74, 1.17
FFS-W 1.02 0.91, 1.13
FFS-S 1.10 0.99, 1.22
HMO-W1 1.17 0.97, 1.41
*** p<0.001

Acute Phase Episode Analyses

Proportions Receiving Services Consistent with Quality Measures in Acute Phase Episodes

Table IV.L.9 describes the proportion of MDD acute phases that occurred either pre- or post-parity for each plan. (NB: Percent pre-parity plus percent post-parity add to 100 percent.)

Table IV.L.9. Proportion of MDD acute phase episodes in pre- and post-parity periods (actual percentages)

Plan Pre-parity Post-parity
MDD Actual Phase Episodes Percent MDD Actual Phase Episodes Percent
FFS-MA1 499 45.4% 601 54.6%
FFS-MA2 440 45.5% 528 54.6%
FFS-NE1 122 37.2% 206 62.8%
FFS-NE2 85 40.5% 125 59.5%
FFS-W 267 39.5% 409 60.5%
FFS-S 340 46.8% 386 53.2%
HMO-W1 124 41.8% 173 58.3%

Tables IV.L.10 through IV.L.15 describe the actual proportions of acute phase episodes, pre- and post-parity, that included the acute phase episode quality measures. One notable trend is that on average, only about 50% to 60% of the episodes experienced the full four months of acute phase follow-up (e.g., follow-up defined as duration of mental health visits or medications). During an episode of treatment, individuals received a minimally appropriate intensity of follow-up between 15.3 percent and 36.5 percent of the time.

For the measures that are conditional upon receiving a certain treatment (i.e., duration/intensity of psychotherapy conditional on receiving any psychotherapy and duration of antidepressant prescriptions conditional on receiving any antidepressants), we provide proportions based on two denominators. In the column “% of MDD sample,” the denominator is all MDD acute phase episodes; in the “% of users” column, the denominator is the total number of acute phase episodes receiving at least one of the treatments described by the specific measure. In the description of these tables, we will focus on the column that is conditional on use (i.e., “% of users”).

The range of use varied across the quality measures by plan. Across plans, similar proportions of acute phase episodes met the minimum quality standard for duration of psychotherapy and antidepressants, typically ranging from less than half to approximately two-thirds of the episodes. The proportion of episodes that met the minimum quality standard for therapy intensity tended to be lower, approximately 20% to 40%.

Table IV.L.10. Proportion of MDD acute phase episodes that met quality measures for duration of follow-up (antidepressants and visits) for the 4 month period

Plan Pre-parity Post-parity
MDD Actual Phase Episodes Met Quality Measures MDD Actual Phase Episodes Met Quality Measures
FFS-MA1 285 57.1% 363 60.4%
FFS-MA2 230 52.3% 311 58.9%
FFS-NE1 60 49.2% 112 54.4%
FFS-NE2 43 50.6% 83 66.4%
FFS-W 144 53.9% 261 63.8%
FFS-S 147 43.2% 205 53.1%
HMO-W1 72 58.1% 100 57.8%

 

Table IV.L.11. Proportion of MDD acute phase episodes that met quality measures for intensity of follow-up (any MH/SA visit) 1st 2months at least 2 per month

Plan Pre-parity Post-parity
MDD Actual Phase Episodes Met Quality Measures MDD Actual Phase Episodes Met Quality Measures
FFS-MA1 169 339.% 182 30.3%
FFS-MA2 110 25.0% 145 27.5%
FFS-NE1 23 18.9% 50 24.3%
FFS-NE2 31 36.5% 52 41.6%
FFS-W 66 24.7% 140 34.2%
FFS-S 52 15.3% 66 17.1%
HMO-W1 36 29.0% 55 31.8%

 

Table IV.L.12. Proportion of MDD acute phase episodes that met quality measures for intensity of follow-up (any MHSA visit) 2nd 2 months at least 1 visit per month

Plan Pre-parity Post-parity
MDD Actual Phase Episodes Met Quality Measures MDD Actual Phase Episodes Met Quality Measures
FFS-MA1 194 38.9% 199 33.1%
FFS-MA2 123 28.0% 152 28.8%
FFS-NE1 39 32.0% 75 36.4%
FFS-NE2 36 42.4% 55 44.0%
FFS-W 85 31.8% 156 38.1%
FFS-S 56 16.5% 86 22.3%
HMO-W1 42 33.9% 64 37.0%

Proportions Receiving Quality Measures in Acute Phase Episodes, Conditional on Use of Specific Treatments

Table IV.L.13. Proportion of MDD acute phase episodes that met quality measures for duration of psychotherapy (individual, group, and family) for the first 3 months

Plan Pre-parity Post-parity
MDD Actual Phase Episodes Met Quality Measures MDD Actual Phase Episodes Met Quality Measures
% of MDD Sample % of users % of MDD Sample % of users
FFS-MA1 187 37.5% 62.8% 209 34.8% 64.1%
FFS-MA2 111 25.2% 51.6% 157 29.7% 54.0%
FFS-NE1 41 33.6% 58.6% 65 31.6% 53.7%
FFS-NE2 41 48.2% 70.7% 50 40.0% 65.8%
FFS-W 77 28.8% 57.5% 162 39.6% 61.6%
FFS-S 62 18.2% 44.9% 100 25.9% 55.0%
HMO-W1 41 33.1% 58.6% 55 31.8% 54.5%

 

Table IV.L.14. Proportion of MDD acute phase episodes that met quality measures for intensity of psychotherapy (individual, group or family) at least 2 per month

Plan Pre-parity Post-parity
MDD Actual Phase Episodes Met Quality Measures MDD Actual Phase Episodes Met Quality Measures
% of MDD Sample % of users % of MDD Sample % of users
FFS-MA1 110 22.0% 36.9% 110 18.3% 33.7%
FFS-MA2 57 13.0% 26.5% 66 12.5% 22.7%
FFS-NE1 15 12.3% 21.4% 25 12.1% 20.7%
FFS-NE2 25 29.4% 43.1% 31 24.8% 40.8%
FFS-W 44 16.5% 32.8% 77 18.8% 29.3%
FFS-S 27 7.9% 19.6% 35 9.1% 19.2%
HMO-W1 15 12.1% 21.4% 28 16.2% 27.7%

 

Table IV.L.15. Proportion of MDD acute phase episodes that met quality measures for cumulative antidepressant duration at least 3 months

Plan Pre-parity Post-parity
MDD Actual Phase Episodes Met Quality Measures MDD Actual Phase Episodes Met Quality Measures
% of MDD Sample % of users % of MDD Sample % of users
FFS-MA1 107 21.4% 56.3% 148 24.6% 63.3%
FFS-MA2 93 21.1% 58.1% 114 21.6% 60.3%
FFS-NE1 22 18.0% 46.8% 34 16.5% 54.8%
FFS-NE2 10 11.8% 41.7% 24 19.2% 53.3%
FFS-W 65 24.3% 67.0% 73 17.9% 57.0%
FFS-S 66 19.4% 54.1% 66 17.1% 52.4%
HMO-W1 30 24.2% 52.6% 30 17.3% 46.2%

Regression Analyses for Quality Measures in Acute Phase Episodes

Table IV.L.16 details the results of the regression analyses for the acute phase episodes. The models controlled for age, gender, relationship to contract holder, and the presence of a co-occurring mental health condition. We were unable to control for co-occurring SA disorder because of a low detected prevalence in the acute phase episode analysis. There were no observed quality decrements associated with the post-parity period. The strongest improvements were seen in the duration of follow-up (FFS-W, FFS-NE2, and FFS-S). There was some improvement in the intensity of follow-up measures for FFS-W and FFS-S. The post-parity period was not associated with any changes in the likelihood of receiving a minimum duration or intensity of psychotherapy; nor was it associated with receiving antidepressants prescription for at least three months duration. It is important to note that in several plans, the sample size of persons receiving any psychotherapy or any antidepressants was too small for adequate power to detect a difference of 10 percentage points based on the minimally acceptable statistical standards of a significance level at 0.05% and power. For the psychotherapy measures, we were underpowered in the FFS-NE2; for the antidepressant duration measure, we were underpowered in HMO-W1 and FFS-NE1, FFS-NE2, and FFS-S.

However, the effectively cross-sectional, rather than longitudinal study samples appear to have had an adverse impact on our ability to detect changes post-parity. Recall that in cross-sectional analyses, persons cannot serve as their own controls which can introduce additional variation, thereby making it more difficult to detect significant change. In order to determine if possibly different results might have been obtained using a longitudinal sample, we performed additional analyses on the small sub-sample of enrollees who received MDD acute phase treatment in both the pre- and post-parity implementation periods (and thus these enrollees were able to serve as their own controls) for some of the plans. In these analyses, we observed positive changes with smaller standard errors that did not achieve significance. These analyses indicate that had there been a larger sample size of enrollees receiving acute phase MDD treatment in both time periods, statistically significant improvements would have been more likely detected.

Table IV.L.16. Regression analysis for odds of receiving acute phase quality measures

Plan Odds ratio Confidence interval
Duration of follow-up (antidepressants & visits) for at least 4 months
FFS-MA1 1.08 0.84, 1.38
FFS-MA2 1.25 0.94, 1.66
FFS-NE1 1.20 0.76, 1.91
FFS-NE2 2.33** 1.31, 4.14
FFS-W 1.72** 1.22, 2.41
FFS-S 1.60** 1.19, 2.16
HMO-W1 0.98 0.62, 1.55
Intensity of follow-up visits (any MH/SA visit) 1st 2 months, at least 2 per month
FFS-MA1 0.84 0.66, 1.07
FFS-MA2 1.09 0.82, 1.43
FFS-NE1 1.49 0.87, 2.56
FFS-NE2 1.01 0.69, 1.49
FFS-W 1.44* 1.04, 2.00
FFS-S 1.13 0.77, 1.67
HMO-W1 1.15 0.68, 1.96
Intensity of follow-up visits (any MH/SA visit) 2nd 2 months, at least 1 per month
FFS-MA1 0.79 0.62, 1.00
FFS-MA2 0.98 0.74, 1.29
FFS-NE1 1.24 0.76, 2.02
FFS-NE2 1.17 0.70, 1.96
FFS-W 1.27 0.91, 1.76
FFS-S 1.49* 1.03, 2.15
HMO-W1 1.22 0.74, 2.01
Conditional on any psychotherapy, duration of psychotherapy for at least 3 months
FFS-MA1 1.07 0.78, 1.48
FFS-MA2 1.06 0.74, 1.50
FFS-NE1 0.76 0.40, 1.41
FFS-NE2 0.95 0.52, 1.78
FFS-W 1.22 0.79, 1.86
FFS-S 1.49 0.95, 2.35
HMO-W1 1.02 0.53, 1.98
Conditional on any psychotherapy, intensity of psychotherapy, at least 2 per month
FFS-MA1 0.84 0.61, 1.15
FFS-MA2 0.83 0.57, 1.21
FFS-NE1 0.95 0.45, 2.00
FFS-NE2 1.01 0.99, 1.02
FFS-W 0.82 0.53, 1.27
FFS-S 1.01 0.55, 1.87
HMO-W1 1.48 0.70, 3.12
Conditinal on any antidepressant prescription, cumulative antidepressant duration for at least 3 months
FFS-MA1 1.31 0.89, 1.94
FFS-MA2 1.01 0.59, 1.72
FFS-NE1 1.38 0.62, 3.09
FFS-NE2 1.35 0.57, 3.23
FFS-W 1.90 0.11, 32.63
FFS-S 0.81 0.47, 1.37
HMO-W1 0.83 0.37, 1.85
* p<0.05
** p<0.01

Summary and Discussion

The post-parity implementation period (years 2001 and 2002) presents a mixed picture in terms of quality improvement.

There was no change in the identification of MDD across plans. Thus, it does not appear in this stable cohort of persons who were enrolled in the insurance plans for all four of the study years that parity resulted in increased diagnostic detection of MDD. Previous literature demonstrates that MDD is under-detected in usual care practice (Kessler, Berglund, Demler, et al., 2003; Simon and Von Korff, 1995)

Both pre- and post-parity implementation, most persons (approximately 90% or better) received at least one of the quality measures in a given person-year. Additionally, in post-parity implementation years, most plans experienced some improvement in some quality indicators, most notably in the likelihood of receiving any antidepressant or any antidepressant or therapy in a given person-year. HMO-W1 is the only plan that did not experience any quality improvement in any of the measures. Also, several plans experienced improvements in the duration of follow-up during acute phase treatment, but there was limited improvement in the intensity of follow-up. There was no improvement observed in the duration/intensity of therapy or duration of antidepressant prescriptions filled. However, we were underpowered to detect a 10 percentage point change in these measures for several plans, particularly for antidepressant duration.

Despite the improvements noted above, there is evidence of quality concerns as well, particularly when one goes beyond the minimal standards of quality in the person-year analyses (i.e., if an enrollee received at least one antidepressant prescription or at least one therapy visit). Whether pre- or post-parity, only 50% to 60% of the acute phases received a minimal duration of follow-up. Also, the intensity of follow-up only met the quality standard about one-third or less of the time. For persons who received antidepressants or psychotherapy, the durations were also typically in a similar range, although there was variation among the plans. The intensity of psychotherapy only met the minimum standard in typically a third or fewer of the plans. The likelihood of receiving some quality measures (e.g., psychotherapy duration/intensity and antidepressant duration) was unchanged post-parity.

Limitations

There are some important limitations to these data. First, these analyses are largely based on Association plans and therefore, the results may not be generalizable to other plans or geographical regions in which parity is implemented. Also, to facilitate comparison, we limit the analyses to enrollees continuously enrolled all four years. Thus, we cannot comment on the association between parity implementation and quality for enrollees who were not enrolled all four years.

The person-year analyses indicate there was considerable improvement in access to any recommended treatments once MDD was identified, but that was a very minimal quality standard (i.e., at least one prescription filled, at least one psychotherapy visit). The more nuanced measures in the acute phase analyses indicate a more modest improvement in terms of magnitude of change and scope across plans. However, additional analyses suggest that had we a larger sample size of enrollees receiving acute phase MDD treatment in both time periods, we may have been more likely to detect statistically significant improvements in the post-parity implementation period.

Additionally, it is important to note that quality improvements in the post-parity implementation period varied by plan and by specific measures. This is likely a result of local contextual differences such as baseline quality in the plans, geographical practice variation (or geographical enrollee preferences), and plan care management and utilization/review strategies.

Finally, these analyses cannot control for secular trends that would affect quality independent of parity implementation. For example, recent literature indicates even prior to 2000 when parity was implemented, MDD has seen increased treatment rates in general, and in particular increased antidepressant utilization, at times while psychotherapy use significantly decreases (Berndt, Frank, and McGuire, 1997; Berndt, Bir, Busch, Frank, and Normand, 2002; Busch, Berndt, and Frank, 2001; Olfson, Marcus, Pincus, et al., 1998; Olfson, Marcus, Druss, et al., 2002). While in these analyses, the strongest improvements were seen in the likelihood of receiving any guideline recommended treatments (i.e., either psychotherapy or antidepressants), typically these gains resulted from increases in the likelihood of receiving antidepressants. Thus, the strongest improvements observed may be entirely related to secular trends, not parity implementation. Difference in difference analyses can adjust for secular trends; such analyses elsewhere in this report examined the probability of MHSA use and spending and found that increases observed in the FEHB plans were similar to trends observed in the comparison data. Thus, it is quite possible that these MDD quality results also reflect secular trends independent of parity implementation.

Impact on Quality of Care for Substance Abuse

Overview and Model

In this section, we report on the analyses of five indicators of quality of substance abuse (SA) treatment in six of the selected FEHB plans.35 These quality measures focus on adult enrollees age 17 to 65 years.36 The first two indicators capture penetration of SA treatment services in the adult enrollee population by examining annual rates of inpatient, residential and outpatient treatment utilization. Utilization rates are reported as the rate of adults using any SA treatment service in the past year, per 1,000 adults who were continuously enrolled in that year.

Among those who received SA treatment services in the past year, we also report on the average intensity of SA-related care. Intensity measures for those who received inpatient or non-hospital residential treatment for SA are number of stays in the past year and average length of stays. Among those who received SA-related outpatient care, the measure of intensity is the average number of outpatient visits in that year.

The next measures describe rates of use of different levels of care (inpatient, residential, and outpatient), and the intensity of care within each level of care. These measures addressed all care within the measurement year, whether or not it represented a new or single episode of care.

These two indicators of SA treatment penetration and intensity are summarized below.

  1. Rate of adults with any inpatient/residential SA care in the measurement year.
    • Among adults with any inpatient SA care, median/mean length of stay.
  2. Rate of adults with any outpatient SA care in the measurement year.
    • Among adults with any outpatient SA care, median/mean number of visits.

The PERT examined three additional quality indicators known as the Washington Circle Measures. We used the definitions and specifications of these measures as they were recently adopted by National Commission on Quality Assurance (NCQA) for the Health Plan Employer Data and Information Set (HEDIS). These measures focus on new, single SA episodes of care. The index visit that defines a new episode of care follows a wash-out period of 60 days of enrollment in which no claims with SA diagnoses are found. The three Washington Circle measures are:

  • the rate of “identification” of an SA diagnosis among adult enrollees;
  • among those identified, the rate of “initiation” of SA treatment; and
  • among those identified, the rate of “engagement” into SA treatment.

The PERT examined the number of cases identified with SA as defined below, using both a one-year and four-year continuous enrollment requirement, to examine its impact on the number of cases available for analysis. Continuous enrollment in a year was required for both the numerator and denominator of calculated rates.

For the one-year continuous enrollment samples, the PERT calculated measures within each of the study years, 1999 and 2000 (pre-parity) and 2001 and 2002 (post-parity). For the four-year, continuous enrollment sample, the PERT calculated measures for the two-year pre-parity period and the two-year post-parity period.

The PERT defined adult enrollees consistent with the rest of the evaluation (must be 18 years of age by entry and less than 65 years at exit). Adult enrollees were analyzed using the Washington Circle measures as defined below:

  1. Identification of adult enrollees with a new SA diagnosis. The rate of identification was calculated as number of adult members with a new SA claim over the measurement period divided by the total continuously enrolled adult members over the measurement period. Adults were defined as enrollees if they were 18 years of age or older by December 31st of the measurement year.
  2. Initiation of SA treatment among adults identified with an SA diagnosis. For this rate, the numerator consisted of adults diagnosed with SA who either had an inpatient/residential SA admission or an outpatient service for SA and an additional SA service within 14 days. The denominator was the number of adults identified with an SA diagnosis (same as the numerator of measure 1).
  3. Engagement in SA treatment among adults identified with an SA diagnosis. This was calculated as a rate; the numerator consisted of adults diagnosed with SA who either had had at least two additional SA services within 30 days after initiation (if initiation visit was an inpatient/residential admission, then the 30-day period began after discharge); the denominator was the number of adults identified with an SA diagnosis (same as the numerator of measure 1).

Specific Analytic Methods for Quality of Care for Substance Abuse

The PERT began with simple descriptive analyses that reported the above rates by plan and measurement year. Note that although the quality measures were defined as rates, the PERT also used the numerators of each rate mentioned above as dependent variables in logistic regression analyses for which the analytic sample is defined by the denominator.

Other Enrollee/Patient Characteristics

The PERT used age to identify adults. For the logistic regression analyses, the PERT included the following variables in the models:

  • Age,
  • Gender,
  • Relationship to health plan contract holder,
  • Any primary or secondary diagnosis of depression during measurement year (definition consistent with rest of the study), and
  • Any other primary or secondary MH diagnosis during the measurement year (definition consistent with rest of the study).

Measure 1: Identification

Relevant Primary and Secondary ICD-9-CM Codes
291-292, 303-304, 305.0, 305.2-305.9, 535.3, 571.1

Priority Rule
If multiple types of services occurred on the same day, then detoxification overrode all other types, and inpatient overrode emergency room (ER) and outpatient. The first service in which SA was identified must have been preceded by a “wash out” period of 60 days in which no other SA related claims occurred.

Identifying SA-related Service Claims

  1. Outpatient visits: Only outpatient visits that were not tests were coded as outpatient visits. Outpatient visits were identified by codes in these groups:
    • Miscellaneous visits;
    • Preventive medicine, evaluation and management, or counseling;
    • Individual therapy;
    • Family therapy; and
    • Group therapy.

    A service was considered an SA service if the procedure was a specific SA procedure or if it had an SA diagnosis (either primary or secondary). Moreover, for outpatient claims in these categories, if the procedure was missing but the principal diagnosis was SA, it was considered as SA service.

  2. ER services: ER codes (if they did not lead to hospitalization) were identified as an SA service in the ER according to the same criteria used for outpatient visits (above).
  3. Detoxification services: All detoxification services were considered specialty SA services.
  4. Inpatient services: Admissions were defined as SA admissions if either the primary or secondary diagnosis was SA. All continuous inpatient days were counted as one admission if no break in the dates occurred, regardless of movement between hospitals or type of beds. Inpatient included both hospital and non-hospital (SA residential facility) care.

Measure 2: Initiation

A patient was considered to have initiated SA treatment if any one of the following three criteria was met:

  1. The patient had an inpatient (non-detoxification) stay with a primary diagnosis of SA and no other SA claims within 60 days before the admit date of the inpatient stay. In this case, the Start Date for the Measure 3 analysis was the discharge date for the inpatient stay.
  2. The patient had an outpatient or ER visit or a missing procedure code with a primary or secondary diagnosis of SA followed within 14 days by a second outpatient (non-ER, non-detoxification) visit or an inpatient (non-detoxification) stay with a primary or secondary diagnosis of SA and no other SA claims within 60 days before the date of the first visit. In this case, the Start Date for the Measure 3 analysis was the date of the second outpatient visit of the discharge date of the inpatient admission.
  3. The patient had a detoxification claim (can be inpatient or outpatient) with a primary or secondary diagnosis of SA or an inpatient stay with SA as a secondary (not primary) diagnosis followed within 14 days after the discharge date by an outpatient (non-ER, non-detoxification) visit with a primary or secondary diagnosis of SA or another inpatient (non-detoxification) stay with SA as a primary or secondary diagnosis. Other SA claims must not have occurred within 60 days before the admit date of the first stay. In this case, the Start Date for the Measure 3 analysis was the outpatient visit date or the inpatient discharge date of the second visit or discharge dated of the inpatient stay.

Measure 3: Engagement

Among patients who had initiated SA treatment, the PERT determined whether another SA service occurred with a beginning date that was within 30 days of the Start Date determined in Measure 2.

The subsequent SA service could be an outpatient (non-ER, non-detoxification) visit with a primary or secondary diagnosis of SA or another inpatient (non-detoxification) stay with SA as a primary or secondary diagnosis. The Washington Circle measures exclude detoxification services from counting as engagement in SA treatment based on evidence that detoxification alone without further rehabilitation-oriented services are not effective in treating SA disorders.

Findings on Quality of Care for Substance Abuse

In this section, we report findings on the impact of the parity policy on five indicators of quality of SA treatment in six FEHB plans. The first two of these quality indicators, presented in Tables IV.M.1 and IV.M.2, are annual rates of inpatient or residential and outpatient treatment utilization per 1,000 continuously enrolled adult enrollees, respectively.

Table IV.M.1. Unadjusted rates and 95% confidence intervals for adults with any inpatient or residential SA treatment, per 1,000 continuously enrolled adults by plan and measurement year

Plan Pre-parity Post-parity
1999 2000 2001 2002
Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval
FFS-MA1 2.6 (2.3, 2.9) 3.0 (2.7, 3.2) 2.6 (2.4, 2.9) 2.8 (2.6, 3.1)
FFS-MA2 2.5 (2.2, 2.8) 2.3 (2.0, 2.6) 2.5 (2.2, 2.8) 2.8 (2.5, 3.1)
FFS-NE1 1.7 (1.4, 2.1) 2.4 (2.0, 2.8) 2.1 (1.8, 2.5) 2.8 (2.4, 3.3)
FFS-NE2 2.4 (1.8, 2.9) 2.2 (1.7, 2.7) 2.6 (2.0, 3.1) 2.2 (1.7, 2.7)
FFS-W 2.5 (2.2, 2.9) 2.1 (1.8, 2.5) 2.9 (2.5, 3.3) 2.8 (2.5, 3.2)
FFS-S 2.2 (1.9, 2.5) 2.5 (2.2, 2.8) 2.4 (2.1, 2.7) 3.1 (2.8, 3.4)

 

Table IV.M.2. Unadjusted rate and 95% confidence intervals for adults with any outpatient SA treatment, per 1,000 continuously enrolled adults by plan and measurement year

Plan Pre-parity Post-parity
1999 2000 2001 2002
Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval
FFS-MA1 3.5 (3.2, 3.8) 4.6 (4.2, 4.9) 4.3 (4.0, 4.6) 5.5 (5.1, 5.9)
FFS-MA2 4.0 (3.7, 4.4) 3.8 (3.4, 4.1) 5.0 (4.6, 5.4) 5.5 (5.0, 5.9)
FFS-NE1 4.0 (3.4, 4.5) 4.8 (4.3, 5.4) 4.5 (3.9, 5.0) 6.5 (5.9, 7.2)
FFS-NE2 3.9 (3.2, 4.6) 3.8 (3.2, 4.5) 5.2 (4.4, 6.0) 6.0 (5.2, 6.8)
FFS-W 3.5 (3.1, 3.9) 3.9 (3.4, 4.3) 4.7 (4.2, 5.2) 5.5 (4.9, 6.0)
FFS-S 3.2 (2.8, 3.5) 3.9 (3.5, 4.2) 4.1 (3.7, 4.5) 4.9 (4.5, 5.3)

Table IV.M.1 presents annual rates (per 1,000 adult enrollees) of SA inpatient or residential treatment utilization. The numerator for these rates represents a count of all continuously enrolled adults receiving inpatient or residential SA treatment in each plan during the measurement year. The denominator represents the total number of continuously enrolled adult enrollees in each plan during the measurement year. Rates were multiplied by 1,000 to obtain rates per 1,000 continuously enrolled adults. For all plans, these rates remained fairly constant over time during the two years pre-parity and two years post-parity with minor fluctuations most likely due to random variation. Three plans (FFS-NE1, FFS-W, and FFS-S) had appreciable increases from pre- to post-parity in rates of inpatient or residential treatment utilization. However, these increases appear to be part of an overall secular trend in these plans.

Table IV.M.2 presents annual rates (per 1,000 adult enrollees) of SA outpatient treatment utilization. The numerator for these rates represents a count of continuously enrolled adult with an outpatient SA visit in each plan during the measurement year. The denominator represents the total number of continuously enrolled adults in each plan during the measurement year. Rates were multiplied by 1,000 to obtain rates per 1,000 adult enrollees.

Table IV.M.2 shows a clear trend of increased outpatient SA treatment in all of the plans during the 4 years of observation. These rate increases began before parity was enacted and continued in the two years after parity. Therefore, there is evidence to suggest that these increases may be due to secular trends rather than parity. It is not possible, however, with the unadjusted rates presented in Table IV.M.2, to tell the level of influence parity had (if any) on the overall magnitude of these rate increases.

Tables IV.M.3 and IV.M.4 break down these indicators into mean numbers of visits and, for the indicator of inpatient or residential service utilization, length of stay.

Table IV.M.3. Mean number of inpatient stays and mean length of stay (LOS) among continuously enrolled adults with any inpatient SA care by plan and measurement year

Plan Pre-parity Post-parity
1999 2000 2001 2002
No. of Stays LOS No. of Stays LOS No. of Stays LOS No. of Stays LOS
FFS-MA1 1.2 10.7 1.3 12.4 1.2 11.7 1.3 11.2
FFS-MA2 1.2 11.7 1.1 8.7 1.2 9.9 1.2 9.4
FFS-NE1 1.2 1.4 1.1 10.1 1.1 8.5 1.4 12.0
FFS-NE2 1.5 14.4 1.3 12.3 1.4 11.8 1.6 13.8
FFS-W 1.2 9.4 1.2 10.3 1.3 12.5 1.3 12.7
FFS-S 1.1 9.8 1.2 9.4 1.2 10.9 1.2 9.8

Table IV.M.3 displays the inpatient and residential treatment utilization data differently than in Table IV.M.1 by presenting the mean number of inpatient stays and mean length of stay of continuously enrolled adults who received inpatient or residential treatment for each plan annually. Therefore, the only members included in these analyses are those members who had inpatient or residential treatment during the measurement year.

All of the means for number of inpatient stays and length of stay, presented in Table IV.M.3, remained relatively constant with little or no fluctuations when comparing the pre-parity and post-parity time periods. The average number of stays in pre-parity years (1999 and 2000) and post-parity years (2001 and 2002) did not differ significantly nor did the average length of stay (data not shown in table). Therefore the data in Table IV.M.3 provide no evidence that the parity policy had any influence on inpatient service utilization in the plans studied.

Table IV.M.4 displays the outpatient treatment utilization data in a different way than in Table IV.M.2 by presenting the mean number of outpatient visits for each plan annually among continuously enrolled adults who received outpatient treatment. Therefore, the only members included in these analyses are those members that had at least one outpatient visit during the measurement year. Table IV.M.4 reveals that annual increases over time in the rate of outpatient SA treatment, as observed in Table IV.M.2, were not matched by any clear increases in the mean number of outpatient visits among patients receiving any outpatient care. In fact, with the possible exceptions of FFS-MA1 and FFS-NE2, the data in Table IV.M.4 do not appear to have any noticeable trends.

Table IV.M.4. Mean number of outpatient visits among continuously enrolled adults with any outpatient SA care by plan and measurement year

Plan Pre-parity Post-parity
1999 2000 2001 2002
FFS-MA1 4.1 6.6 4.6 7.3
FFS-MA2 4.1 4.5 4.3 4.8
FFS-NE1 3.7 6.1 3.1 3.8
FFS-NE2 8.6 8.8 9.8 9.9
FFS-W 3.6 3.5 3.8 3.7
FFS-S 2.7 3.8 3.7 3.4

Tables IV.M.5, IV.M.6 and IV.M.7 present the unadjusted annual rates (per 100 adult enrollees identified with an SA problem) of the remaining three quality indicators which were developed by the Washington Circle Group: identification of adult enrollees with a new SA disorder, initiation of SA treatment among adults with a primary or secondary SA diagnosis, and engagement of SA treatment among adult enrollees with a primary or secondary SA diagnosis, respectively.

Table IV.M.5 presents unadjusted SA identification rates of adults with a new SA diagnosis per 1,000 continuously enrolled adults. The numerator for these rates represents a count of continuously enrolled adults with a new primary or secondary SA diagnosis during the measurement year as identified by an index visit. An index visit defines a new episode of care and has to follow a washout period of 60-days of enrollment in which no claims with SA diagnoses are found. The denominator represents the total number of continuously enrolled adults in each plan during the measurement year. Rates were multiplied by 1,000 to obtain rates per 1,000 adult enrollees.

Table IV.M.5. Unadjusted SA identification rates and 95% confidence intervals among adults with a new SA diagnosis, per 1,000 adult enrollees by plan and measurement year

Plan Pre-parity Post-parity
1999 2000 2001 2002
Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval
FFS-MA1 4.5 (4.1, 4.8) 5.1 (4.7, 5.4) 4.9 (4.5, 5.2) 5.7 (5.3, 6.1)
FFS-MA2 4.7 (4.3, 5.1) 4.5 (4.1, 4.9) 5.3 (4.9, 5.7) 5.9 (5.5, 6.4)
FFS-NE1 4.1 (3.6, 4.6) 4.8 (4.3, 5.4) 4.7 (4.2, 5.3) 7.0 (6.3, 7.6)
FFS-NE2 4.2 (3.5, 4.9) 4.2 (3.5, 5.0) 4.8 (4.1, 5.6) 5.6 (4.8, 6.4)
FFS-W 4.6 (4.1, 5.1) 4.8 (4.3, 5.2) 5.8 (5.3, 6.3) 6.2 (5.7, 6.8)
FFS-S 4.3 (3.9, 4.7) 4.7 (4.3, 5.1) 4.9 (4.5, 5.3) 5.9 (5.5, 6.4)

Table IV.M.5 reveals in the post-parity period, all of the health plans under consideration experienced a small increase in their rates of identifying adults with a new SA diagnosis compared to the pre-parity period. The largest change in rate from the pre-parity period to the post-parity period was 2.9 per 1,000 for FFS-NE1, while the smallest change in rate during the same time period was 1.2 for FFS-MA2 (data not shown in table). The largest changes in rates in identifying adults with a new SA diagnosis occurred in the post-parity time period from 2001 to 2002. It is not possible to tell from these unadjusted rates if these increases represent merely secular changes or instead are an effect of the parity policy.

Table IV.M.6. Unadjusted SA initiation rates and 95% confidence intervals among adults with a new SA diagnosis, per 100 adult enrollees identified with a new SA diagnosis by plan and measurement year

Plan Pre-parity Post-parity
1999 2000 2001 2002
Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval
FFS-MA1 22.0 (18.5, 25.4) 26.7 (23.1, 30.2) 20.8 (17.6, 24.1) 37.9 (33.9, 41.8)
FFS-MA2 26.1 (21.6, 30.5) 23.6 (19.4, 27.9) 24.9 (20.9, 28.9) 24.2 (20.5, 27.9)
FFS-NE1 31.1 (23.9, 38.4) 33.3 (26.6, 40.1) 27.5 (21.3, 33.7) 28.4 (23.3, 33.4)
FFS-NE2 32.1 (22.4, 41.8) 31.9 (22.6, 41.2) 35.4 (26.3, 44.5) 35.7 (27.3, 44.1)
FFS-W 20.8 (16.1, 25.5) 20.8 (16.3, 25.4) 28.4 (23.6, 33.3) 20.2 (16.3, 24.0)
FFS-S 13.4 (10.0, 16.9) 21.8 (17.8, 25.8) 21.8 (17.9, 25.7) 23.7 (20.2, 27.3)

Table IV.M.6 presents unadjusted SA initiation rates among continuously enrolled adults with a new SA diagnosis per 100 adults identified with an SA diagnosis. The numerator for this rate represents a count of all continuously enrolled adults in which SA care was initiated during the membership year. The denominator represents the total number of continuously enrolled adults identified with a new SA diagnosis. Rates were multiplied by 100 to obtain rates per 100 adults identified with an SA diagnosis.

Table IV.M.6 reveals that there were no clear increases in initiation rates (per 100 continuously adult enrollees identified) for any of the plans when comparing pre-parity 1999 and post-parity 2002 rates. Plans fluctuated in their rates from year to year, some trending higher and some trending lower by 2002. No clear change from pre- to post-parity was apparent in these data.

Table IV.M.7. Unadjusted SA engagement rates and 95% confidence intervals among continuously enrolled adults with a new SA diagnosis, per 100 adults with a new SA diagnosis by plan and measurement year

Plan Pre-parity Post-parity
1999 2000 2001 2002
Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval Unadj. rate Confidence interval
FFS-MA1 9.1 (6.9, 11.3) 4.0 (11.4, 16.5) 8.5 (6.4, 10.5) 13.7 (11.3, 16.1)
FFS-MA2 13.0 (9.9, 16.2) 11.7 (8.7, 14.7) 11.9 (9.1, 14.6) 10.6 (8.2, 13.0)
FFS-NE1 10.9 (6.0, 14.2) 18.3 (13.3, 23.3) 7.7 (4.4, 11.0) 13.9 (10.4, 17.5)
FFS-NE2 19.8 (12.2, 27.5) 20.6 (13.1, 28.1) 20.1 (13.3, 27.0) 23.0 (16.3, 29.7)
FFS-W 6.9 (4.2, 9.7) 8.3 (5.4, 11.2) 11.6 (8.5, 14.7) 8.4 (5.9, 10.9)
FFS-S 4.1 (2.2, 6.0) 10.4 (7.7, 13.2) 8.6 (6.1, 11.0) 8.9 (6.7, 11.0)

Table IV.M.7 presents unadjusted SA engagement rates among continuously enrolled adults with a new primary or secondary SA diagnosis per 100 adults identified with an SA diagnosis. The numerator for this rate represents a count of all continuously enrolled adults that engaged in SA treatment during the membership year. The denominator represents the total number of continuously enrolled adults identified with a new SA diagnosis. Rates were multiplied by 100 to obtain rates per 100 continuously enrolled adults identified with an SA diagnosis. Compared to pre-parity 1999, post-parity 2002 engagement rates increased for all but one plan, FFS-MA2, which remained the same. There was, however, considerable fluctuation and random variation in these rates over the four years. For example, three plans’ rates increased from 1999 to 2000, then decreased from 2000 to 2001, then increased again from 2001 to 2002. Thus, no clear pre- to post-parity change in these rates over time was evident, despite the fact that all but one plan experienced increases in their engagement rates.

Table IV.M.8 presents multivariate unconditional logistic regression analyses of these three quality indicators. Standard errors for these models were calculated using a GEE approach. These models allowed us to examine the relationship between the parity policy and SA quality indicators. Three separate models were run for each plan. The primary outcome variables for each of these models were the three quality indicators created by the Washington Circle Group.

The primary predictor variable for these models was a binary variable representing pre-parity (1999 and 2000) and post-parity (2001 and 2002) time periods. All of these models controlled for age, gender, mental health diagnosis, relationship to contract holder, and time trends. The only quality indicator significantly associated with the parity policy was identification. All of the plans except one (FFS-NE2) had significant although modest increases in identification during the post-parity period compared to the pre-parity period. Neither initiation nor engagement was significantly associated with implementation of the parity policy after adjusting for covariates.

Table IV.M.8. Multivariate unconditional logistic regression analyses comparing three SA quality indicators, pre-parity and post-parity†

Plan Identification Initiation Engagement
OR 95% CI OR 95% CI OR 95% CI
FFS-MA1 1.04 (1.00, 1.07)* 1.00 (0.92, 1.09) 0.98 (0.85, 1.14)
FFS-MA2 1.07 (1.03, 1.12)*** 0.99 (0.90, 1.09) 0.94 (0.79, 1.12)
FFS-NE1 1.13 (1.07, 1.19)*** 0.92 (0.81, 1.05) 0.89 (0.71, 1.11)
FFS-NE2 1.08 (1.00, 1.17) 1.10 (0.93, 1.31) 0.98 (0.73, 1.32)
FFS-W 1.08 (1.03, 1.14)** 1.10 (0.97, 1.24) 1.19 (0.99, 1.43)
FFS-S 1.05 (1.01, 1.09)* 1.11 (1.00, 1.23) 0.94 (0.76, 1.16)
† Controlling for age, gender, mental health diagnosis, relationship to contract holder, and time trends.
* p<0.05
** p<0.01
*** p<0.001

Discussion

The analyses presented in this section detailed the relationship between the FEHB parity policy and five SA care quality indicators. With the exception of modest increases in identification rates over time, there was no strong evidence to suggest that the parity policy had any influence on these quality indicators. Having data from a number of large health plans across the country and a sound analytic strategy strengthened the analyses presented in this section. The analyses, however, also have a number of limitations.

First, the results in Table IV.M.8 are based on a series of 18 separate logistic regression analyses (three for each of the six plans considered). A conservative approach would be to adjust the significance level for these tests because of multiple comparisons using an approach by Bonferroni, Scheffe or Tukey. Such an adjustment, though, would nullify the significant association between SA identification and the parity policy.

Second, the sample size for these analyses was limited to the small number of SA service users in each of the six plans who were receiving SA treatment. The small size of the sample decreases the precision of the confidence interval calculations and limits our ability to infer conclusions from the data. Sampling a larger number of plans and having a larger sample size would increase our confidence in the findings.

Third, all six plans in these analyses were fee-for-service plans. The generalizability of these findings beyond fee-for-service plans may be limited. In addition, all of the geographic regions represented by the seven plans are large. It is unclear whether these findings would generalize to smaller regions.

View full report

Preview
Download

"parity.pdf" (pdf, 5.07Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"parityA.pdf" (pdf, 61.31Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"parityB.pdf" (pdf, 194.85Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"parityC.pdf" (pdf, 134.89Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®

View full report

Preview
Download

"parityD.pdf" (pdf, 147.52Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®