PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Psychiatr Serv. Author manuscript; available in PMC 2014 May 2.
Published in final edited form as:
PMCID: PMC4007069
NIHMSID: NIHMS574586

A comparison of phone-based and onsite-based fidelity for Assertive Community Treatment in Indiana

Abstract

Objective

This study investigated the reliability, validity, and role of rater expertise in phone-administered fidelity assessment instrument based on the Dartmouth Assertive Community Treatment Scale (DACTS).

Methods

An experienced rater paired with a research assistant without fidelity assessment experience or a consultant familiar with the treatment site conducted phone based assessments of 23 teams providing assertive community treatment in Indiana. Using the DACTS, consultants conducted on-site evaluations of the programs.

Results

The pairs of phone raters revealed high levels of consistency [intraclass correlation coefficient (ICC)=.92] and consensus (mean absolute difference of .07). Phone and on-site assessment showed strong agreement (ICC=.87) and consensus (mean absolute difference of .07) and agreed within .1 scale point, or 2% of the scoring range, for 83% of sites and within .15 scale point for 91% of sites. Results were unaffected by the expertise level of the rater.

Conclusions

Phone based assessment could help agencies monitor faithful implementation of evidence-based practices.

Poor implementation of evidence based practices (1, 2) and its toll on program outcomes are a critical concern for mental health services (3, 4). One accepted strategy to improve implementation is to verify program fidelity (5, 6). However, with the increasing number of evidence based practices, the need to conduct fidelity measurement has begun to place a very high burden on agencies charged with ensuring service quality. For example, the current standard fidelity instrument for ACT, the Dartmouth Assertive Community Treatment Scale (DACTS; 7), requires one day for the onsite visit and another day to score and write the report for quality improvement feedback.

In response to these and related problems, a 2007 national task force met to identify alternative approaches for ensuring quality (8). Among the strategies discussed was alternative fidelity methods such as phone-administered assessments. Although phone-administered fidelity has been used successfully in predicting outcomes (9), no research has validated phone-administered assessment compared to onsite assessment. We examined the interrater reliability and concurrent validity of phone-administered fidelity as applied to Assertive Community Treatment. A secondary question asked whether validity and reliability are higher when using raters with prior experience with fidelity assessment or experience with the site.

Methods

Thirty-two ACT teams in Indiana were invited to participate; 23 (71.9%) agreed. All programs had been operating for at least one year, adhered to the Indiana ACT standards, and were receiving annual onsite fidelity assessments from the ACT Center of Indiana (10). The study took place between October 2008 and March 2010.

Onsite Fidelity

The 28-item DACTS (11) assesses fidelity to ACT along three dimensions: Human Resources (e.g., psychiatrist on staff), Organizational Boundaries (e.g., explicit admission criteria), and Nature of Services (e.g., in-vivo services). Items are rated using a 5-point behaviorally-anchored scale (5=full implementation, 1=not implemented). Mean item scores of 4 and above are considered characteristic of established ACT teams. The DACTS has excellent interrater reliability (11) and can differentiate between ACT and other types of intensive case management (7).

Phone-based Fidelity

A phone protocol was developed based on prior phone-based assessment experience (13), and incorporated two key principles. (1) Convert subjective, global questions into molecular, objective data, e.g., replace a global evaluation of responsibility for treatment with a review of specific clients receiving services outside the ACT team. (2) Use of a tabular format, e.g., the staffing table included information about role and hours on team, supervisor, team meeting attendance, turnover, and vacancies, providing data for scoring 11 DACTS items. The DACTS phone protocol included nine tables (plus detailed instructions for completion): staffing, caseload and discharges, client admissions, client hospitalizations, client contact hours and frequency, services received outside of ACT, engagement mechanisms, substance abuse treatment, and miscellaneous (program meeting, practicing team leader, crisis services, and work with informal supports).

Fidelity Assessment Experience

Participants answered the following questions about the phone assessment process: time to complete and prepare and open-ended questions relating to preparation activities, assessment burden/helpfulness, and suggestions for improvement.

Phone interview

To insure that information on burden was not confounded by prior completion of the onsite visit, DACTS phone interviews were conducted before (M=6.78 days), but no more than one month prior to, the onsite visit. However, due to scheduling difficulties, the onsite interview occurred 49 days after the phone interview at one site and 12 days before the phone interview at another. Sites received a copy of the phone protocol for review two weeks before the phone interview. Team leaders consulted clinical and other program records to complete tables and were encouraged to contact the research team with questions.

Phone interviews were conducted by two raters. To assess impact of rater expertness and prior onsite experience, we systematically varied raters. For half of the interviews, raters were not directly involved with training/consultation of ACT teams: the first author (experienced rater), who has extensive experience conducting both phone and onsite fidelity assessment and a research assistant with no prior experience with fidelity assessment (naïve rater). The remaining interviews were conducted by the first author and the assigned ACT Center consultant, who was familiar with the site. Sites were assigned a second rater (consultant or naive) using quota sampling, stratified by population density (rural vs. urban) and consultant (there were four consultants). Assignment was balanced across the two strata. Adjustments were required when teams declined to participate or had scheduling conflicts. Overall, consultants rated about half of their teams (57%, 33%, 43%, and 50%) and the naïve rater conducted calls with about half of the urban (53%) and rural teams (50%).

To test the validity of phone interviews with minimal staff burden, the team leader was the only site participant. The interview focused on reviewing completed tables for accuracy. In the three cases where the tables were not completed before the call, the interview focused on completing the tables together with the team leader.

Raters independently scored the fidelity items and then discussed their scores to come to consensus. Research team raters based their scores solely on the answers given during the phone interview. Consultants’ scores could be informed by knowledge of ACT team operation from prior contact.

Onsite interview

The ACT Center consultant assigned to the team conducted the onsite fidelity assessments. Consultants mailed a checklist of fidelity assessment items/activities to team leaders prior to the site visit (e.g., team roster, interviews with specific staff members). The phone protocol was not used for the onsite interview. The onsite visit typically involved a one hour observation of the daily team meeting, 1.5 – 2 hours to interview the program leader, a half-hour interview with the substance abuse specialist, 1–2 hours shadowing team members in the community and interviewing clients, 2–3 hours of chart and other record reviews, and a half-hour for wrap-up questions with the program leader. Consultants completed DACTS scoring within 5 working days of the site visit and were free to contact program leaders to clarify data if needed.

Consultants received extensive initial training on the DACTS, reviewed the DACTS protocol and scoring at an all-day training workshop annually, and had at least two years experience conducting DACTS assessments. Questions/issues regarding DACTS scoring were addressed through email contact and a bi-weekly meeting.

Although we could not verify onsite reliability, during the first three years of the contract and throughout Indiana’s participation in the National Implementing Evidence-based Practices Project, all onsite assessments were completed using two raters, with nearly perfect interrater reliability, intraclass correlation (ICC)=.99 (15).

Analysis

We adopted Stemler’s (12) suggestion to use both consensus (raters agree closely and adopt common meaning of scale) and consistency estimates (raters rank sites similarly and are self consistent in their application and understanding of scale) of interrater reliability. Consistency may be high when consensus is low. For phone interviews, interrater consistency was calculated using the ICC. Interrater reliability was calculated across all rater pairs (experienced vs. second rater) and separately for experienced vs. consultant and experienced vs. naïve rater. Interrater consensus was indexed using the mean and range of the absolute value of the difference between raters. Concurrent validity between phone and onsite ratings was calculated using ICCs (consistency) and the mean and range of the absolute value of the difference between ratings (consensus). Phone and onsite interview scores were compared for DACTS total and subscale scores and when using different raters/ rater pairs (i.e., consensus, naïve rater, consultant rater, phone experienced rater). Calculations for all ICCs followed Shrout and Fleiss (13) Model 2 and used two-way random effects ANOVA with absolute agreement. ICCs above .90 are very good, above .80 are acceptable, and above .70 are adequate for exploratory research (14).

Results

Is phone-based fidelity assessment reliable?

Focusing first on the experienced vs. second rater comparisons, the ICCs indicated high levels of reliability (consistency agreement) for Total (ICC=.92), Human Resources (ICC= .93) and Nature of Services (sub)scales (ICC=.91), and adequate reliability for Organizational Boundaries (ICC=.78) (see Table 1). Absolute differences between raters also were small, indicating consensus, for Total, Organizational, and Human Resources subscales using both the mean (.07, .08, .11) and the range (largest discrepancy < .3). Absolute differences for Nature of Services were slightly larger (mean=.18), with the largest discrepancy of .50. There was no discernable impact of prior experience with phone assessment or with the site on consistency (ICCs ranged from .91 to .92) or consensus agreement (mean differences ranged from .06 to .07).

Table 1
Dartmouth Assertive Community Treatment Scale (DACTS) total scale and subscale reliability and validity

Are phone-based fidelity ratings valid?

The onsite and phone-based ratings demonstrated consistency and consensus (Table 1). Focusing first on consensus ratings, ICCs indicated strong agreement between phone and onsite (consistency) for the Total (ICC=.87), Human Resources (ICC= .88) and Nature of Services subscales (ICC=.87), and lower agreement for Organizational Boundaries (ICC=.69). With one exception, the absolute differences between phone and onsite (consensus) tended to be small for the scales/subscales as measured using the mean (mean differences < .14) and the range (largest discrepancy < .33). However, Nature of Services had a discrepancy between phone and onsite at one site of .5. Phone and onsite ratings for the Total scale differed by .10 or less for 19 sites (83%) and by .15 points or less for 21 sites (91%).

There was a small effect of phone rater on consistency (ICC) but not consensus (Table 1). The ICC between phone and onsite was highest when the consultant did both ratings (.92), was similar to the consensus rating when phone ratings were made by the experienced rater (ICC=.86) and lowest when made by the naïve rater (ICC=.79).

Fidelity phone calls averaged 71.5 +/− 20.5, and ranged from 40 to 111 minutes. Site preparation time for the phone interview, averaged a workday (M=7.6+/−5.9 hours) and ranged from 1.8 hours to 25.0 hours. Preparation time was impacted by availability of electronic medical records and variability in record keeping (e.g., ongoing tracking of clinical activities). Universally, team leaders liked the phone assessment, particularly the table format, felt it was straightforward and rated it either less difficult or comparable to preparing for onsite assessment. However, they expressed concerns that phone assessment should not be the exclusive method of fidelity assessment, worried that it limits contact with consultants, reducing training opportunities and ecological validity of assessment, and suggested including other team members during the assessment, especially the substance abuse specialist.

Discussion

The results indicate that phone assessment is reliable and valid. Phone assessment also appeared to be unbiased, i.e., neither over- or under-estimating onsite scores, and accurate, agreeing with the onsite assessment within .1 scale point (2% of the scoring range) for 83% of sites and within .15 scale point for 91% of sites. These results provide strong support for the usefulness of phone fidelity assessment.

Surprisingly, rater prior experience, either with phone assessment or the site, had no discernable impact on reliability and only a minor and ambiguous impact on validity. Increased phone-onsite consistency for the consultant rater probably reflected method variance (same rater for both assessments), rather than increased accuracy. Moreover, increased consistency for the experienced rater was not matched by increased consensus with onsite scores. Two factors may explain the small impact of rater: the minor role of the interview(er) in the phone assessment process and the success of the phone protocol in creating an objective, molecular format for gathering fidelity data. For example, protocol preparation time averaged nearly a day, whereas the phone interview took about an hour. Because the phone interview largely focused on verifying the information already tabulated by the team leader, the rater’s role during the interview was less of an expert observer, and more of an auditor ensuring self-report accuracy. These results suggest that self-report, if based on clear, objective criteria, may be a useful adjunctive method for fidelity assessment.

The study has several limitations. All the sites were in one state, limiting generalizability, and were certified as ACT teams, limiting the range of fidelity explored. Team leaders were not blind to the study hypotheses, possibly biasing the results, and many were experienced with fidelity reviews, potentially underestimating preparation time required for those naïve to fidelity measurement. Onsite fidelity was conducted using a single rater, providing no opportunity to verify interrater reliability, and the experienced rater was limited to a single individual, limiting generalizability. Finally, we did not measure onsite assessment preparation or site visit time, limiting our ability to compare burden levels.

Despite limitations, the study provides strong evidence for the viability of phone-based ACT fidelity assessment. Further work is needed to examine phone fidelity with other evidence based practices (e.g., supported employment) and with new teams or lower fidelity teams in situations when reimbursement is contingent on high scores. There are several caveats to phone fidelity. For example, although burden to assessors was low, preparation time burden to sites was still high, perhaps prohibitively so. Moreover, phone assessment provides limited opportunity for training and interaction with clients and team members. Thus, phone fidelity cannot and probably should not fully replace onsite fidelity. Instead, both could be integrated into a stepped fidelity assessment approach (15). Onsite fidelity is likely uniquely valuable for assessing teams starting up or experiencing a major transition (e.g., high team turnover). Phone fidelity is likely ideal for stable, mature teams and for frequent check-ins. Future work should explore the relative uses of both methods.

Acknowledgements

This study was funded by an IP-RISP grant from the National Institute of Mental Health (R24 MH074670; Recovery Oriented Assertive Community Treatment). We appreciate the assistance of others in the collection of data for this study.

Footnotes

Disclosures: None for any author.

References

1. Phillips SD, Burns BJ, Edgar ER, et al. Moving assertive community treatment into standard practice. Psychiatric Services. 2001;52:771–779. [PubMed]
2. Drake RE, Essock SM, Shaner A, et al. Implementing dual diagnosis services for clients with severe mental illness. Psychiatric Services. 2001;52:469–476. [PubMed]
3. McHugo GJ, Drake RE, Teague GB, et al. The relationship between model fidelity and client outcomes in the New Hampshire Dual Disorders Study. Psychiatric Services. 1999;50:818–824. [PubMed]
4. McGrew JH, Bond GR, Dietzen LL, et al. Measuring the fidelity of implementation of a mental health program model. Journal of Consulting and Clinical Psychology. 1994;62:670–680. [PubMed]
5. Mancini AD, Moser LL, Whitley R, et al. Assertive community treatment: Facilitators and barriers to implementation in routine mental health settings. Psychiatric Services. 2009;60:189–195. [PubMed]
6. Bond GR, Williams J, Evans L, et al. Psychiatric Rehabilitation Fidelity Toolkit. Cambridge, MA: Human Services Research Institute; 2000.
7. Teague GB, Bond GR, Drake RE. Program fidelity in assertive community treatment: development and use of a measure. American Journal of Orthopsychiatry. 1998;68:216–32. [PubMed]
8. Evidence-based Practice Reporting for Uniform Reporting Service and National Outcome Measures Conference; September 2007; Bethesda, MD.
9. McGrew JH, Griss ME. Concurrent and predictive validity of two scales to assess the fidelity of implementation of supported employment. Psychiatric Rehabilitation Journal. 2005;29:41–47. [PubMed]
10. Salyers MP, McKasson RM, Bond GR, et al. The role of technical assistance centers in implementing evidence-based practices: lessons learned. American Journal of Psychiatric Rehabilitation. 2007;10:85–101.
11. McHugo GJ, Drake RE, Whitley R, et al. Fidelity outcomes in the national implementing evidence-based practices project. Psychiatric Services. 2007;58:1279–1284. [PubMed]
12. Stemler SE. A comparison of consensus, consistency, and measurement approaches to estimating interrater reliability. [retrieved June 8, 2010];Practical Assessment, Research & Evaluation. 2004 9 from http://PAREonline.net/getvn.asp?v=9&n=4,
13. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychological Bulletin. 1979;86:420–428. [PubMed]
14. Lombard M, Snyder-Duch J, Bracken CC. Content analysis in mass communication: assessment and reporting of intercoder reliability. Human Communication Research. 2002;28:587–604.
15. McGrew J, Stull L. Alternate methods for fidelity assessment, in Gary Bond Festschrift Conference; September 23, 2009; Indianapolis, IN.