HCAHPS Mode Experiment III

ATTACHMENT C – HCAHPS MODE EXPERIMENT III

Objectives

1. For the new Coordination of Care (CoC) composite measure, analyze Mode

Experiment III results for the following:

1. Patient-mix adjustment to inform how patient sub-groups respond to the new measure.

2. Potential mode effects for the CoC measure.

3. Describe how CoC relates to other measures using psychometric analyses.

2. Analyze the potential effectiveness of new patient-mix adjustment survey items

related to self-rated mental health and emergency room admission.

Table of Contents

Appendices:
Appendix A.1 Descriptive statistics for CoC items and composite, overall and by mode
Appendix A.2 Cross tab of CoC items with Survey Mode
Appendix A.3 Cross tab of CoC items with Survey Mode, missing included
Appendix B.1 Cronbach Alpha and inter-item correlation matrix
Appendix B.2 Correlation of linear coded CoC items and composite with other measures
Appendix B.3 Correlation of top-box coded CoC items and composite with other measures
Appendix C.1 Linear regression models predicting linear coded CoC composite from standard PMA, adding ER admission and/or MHP.
Appendix C.2 Linear regression models predicting top-box coded CoC composite from standard PMA, adding ER admission and/or MHP.
Appendix D Mode effects for CoC linear and top-box composite .
Appendix E Hospital-level ICC, Spearman correlation and reliability.
Appendix F.1 PMA adjusted mean scores for linearly and top-box scored measures Appendix F.2 Correlation and Informativeness of adjusted mean scores from Model 1 with alternative models
Appendix G Difference in adjusted mean scores from standard PMA Model (Base Model) and each of the alternate Models 2, 3 and 4 Appendix H Mode effects, with and without PMA adjustment Appendix I Mixed effect models with mode and standard PMA as fixed effects, and hospital and hospital by mode as random effects.

ii

APPENDIX A

1. Descriptive Statistics

   1. METHODS

We produced tables that present descriptive statistics for age, service line, and all survey items for completed surveys overall and by mode. The descriptive statistics include means and simple frequencies. For each survey item, we produced a version of the descriptive statistics that includes percent missing and a version that excludes the missing cases

2. RESULTS

The mean, standard deviation, and number observed are presented for questions with ordinal response (Appendix A.1). The frequencies, overall and by mode, for the all variables are presented in Appendix A.2 (including percent missing) and A.3 (excluding the missing cases).

After reversing coding, the mean overall health and mental health perceptions on a scale from poor (1) to excellent (5) are 3.15 and 3.77 respectively, which corresponds to good health and very good mental health. More than half (62%) of patients reported having been admitted via the emergency room.

3. CONCLUSIONS

The distributions of HCAHPS items are as expected, with the exception that self-reported emergency room admission rates notably exceeded the rates seen for the administrative measures of emergency room admission, which was nearer to 40%.

APPENDIX B

2. Coordination of Care (CoC)/Transition Items

   1. METHODS

There are three CoC items on a 4-point scale (strongly agree - strongly disagree), but one of the items has a fifth response value (a tailored not applicable) that we treated as missing. We generated two versions of each CoC item: 1) linear mean scoring on a 0-100 scale and b) top-box scoring (top = strongly agree vs. all other except not applicable/missing). We used both the linear and top-box versions of the three CoC items and the CoC composite in all analyses below.

For each of the three CoC items, we generated an inter-item correlation matrix and Cronbach’s Alpha. We also examined the correlation of the CoC composite and each CoC item with each of the 10 reported HCAHPS measures (correlating top-box with top-box and linear with linear).

2. RESULTS

The mean score for the CoC composite, with 0-100 linear mean scoring, is 80, which is between agree and strongly agree. The top-box proportion (strongly agree) for the CoC composite is 47%.

The inter-item correlations with linear scoring are between 0.50 and 0.65, and the standardized Cronbach’s alphas are 0.82 and 0.80 for the linear and top-box CoC composites, respectively (Appendix B.1). The Pearson and Spearman correlations for linear and top-box CoC composite and items with other HCAHPS measures are presented in Appendices B.2 and B.3, respectively. The Pearson correlations for the linearly scored CoC composite with other linearly scored HCAHPS measures are between 0.3 and 0.5, and slightly lower for the three individual CoC items (highest for Q23-staff took preferences into account and lowest for Q25-understood medications). The Spearman correlations show a similar pattern, and overall are lower than the Pearson correlations. The highest correlation with the linearly scored CoC composite is for recommending hospital (r=0.51) followed by Communication with Nurses (r=0.50), Communication about Medicine (0.49) and Rating of Hospital (0.48).

3. CONCLUSIONS

The CoC measure shows no evidence of a ceiling effect. The high Cronbach’s alpha suggests that the CoC measure has very good internal consistency reliability. The moderately high association with hospital rating and recommendation are evidence of validity and suggest that patients value good coordination of care. The moderate correlations with other HCAHPS measures indicate that the CoC composite is not redundant with other HCAHPS measures. The strong association of CoC with Communication with Nurses and Communication about Medicine suggests the importance of coordination of care to those domains.

APPENDIX C

3. Patient-Mix Adjustment (PMA)

   1. METHODS

The purpose of the patient-mix adjustment analyses was to evaluate two new items: ER admission (ERA) and Mental Health Perception (MHP). We compared four sets of models:

1. Base model that corresponds to current official HCAHPS PMA¹ (hospital intercepts plus current PMAs)

2. Alternate model 1: add ERA to base model

3. Alternate model 2: add 1df (linear) MHP to the base model

4. Alternate model 3: add both ERA and MHP to base model

For each patient mix model we ran four versions (one version corresponding to the base model, and three to Alternate Models 1-3) of 22 linear models (11 total outcomes, including the new CoC composite x linear/top-box scoring). We report the coefficients associated with the patient-mix adjusters for all 4 models. The new (candidate) patient-mix adjusters (ERA and MHP) appear on top of the table. We will also examine shifts in current patient-mix adjusters, including general health perception and service line, from the base model to the alternative models (these coefficients appear just below ERA and MHP).

b. RESULTS

For each linear measure the parameter estimates and significance from the four models are presented side-by-side in Appendix C.1. Similarly, the results for the top-box measures are presented in Appendix C.2. Regardless of linear or top-box coding, MHP is significant for all measures in both Models 2 and 3. In all instances the MHP coefficient indicates more positive experiences with better MHP. In contrast, ER admission alone or in combination with MHP is significant for only 3 (doctor communication, discharge info, CoC) of the 11 linear measures and 2 (doctor communication and discharge info) of the 11 top-box measures. In all of these instances the coefficient for ERA was negative, indicating less positive experiences for those admitted through the ER.

c. CONCLUSIONS

The associations of MHP and ERA with HCAHPS items are in the direction expected. Future research will inform the advisability of these items as patient-mix adjustors.

APPENDIX D

4. Mode Analyses

1. METHODS

We estimated a base series of models. Each model predicted CoC outcomes (linear and top-box) from hospital indicator and the three mode indicators (reference=mail mode). This version is not patient-mix adjusted.

2. RESULTS

The coefficients and significance for three mode indicators (CATI, Mixed, and TT-IVR) versus Mail Only mode are presented in Appendix D.

With respect to the CoC composite, the scores for CATI are significantly higher than Mail Alone mode for both linear and top-box scoring. However, we find no significant differences between IVR and Mixed mode relative to Mail Only mode.

3. CONCLUSIONS

The mode effect results for CoC are similar to what has been observed previously in Mode Experiment 1.

APPENDIX E

1. Coordination of Care (CoC)/Transition Items

   1. METHODS

We will calculate the interclass correlation coefficient (ICC) at the hospital-level and Spearman-Brown (hospital-level) reliability of the new potential CoC composite and compare these to the ICCs for the 10 currently reported HCAHPS measures (linear mean and top-box versions of each).

2. RESULTS

Appendix F presents the ICC and the reliability statistics. ICC measures similarity of patients within a hospital and ranges between 0 and 1 (theoretically negative values are possible). The observed ICC’s (linear and top-box coded) across all measures are very low indicating that patients within hospitals are dissimilar with respect to the eleven HCAHPS measures. The reliabilities (at an average 130 completes) range from poor (<0.50) for Pain Management and Communication about Medicines to high (>0.90) for Quiet and Hospital Recommendation. The estimated Spearman-Brown reliabilities (at n=300) are good ranging between 0.71 and 0.92. The ICC and reliability (including Spearman-Brown Reliability) statistics are consistent across scoring methods (i.e., linear scoring and top-box scoring). The new Coordination of Care measure has ICC and reliability (at n=130) above the median for current HCAHPS measures under both linear and top-box scoring.

3. CONCLUSIONS

The new CoC measure has psychometric properties as good or better than current HCAHPS measures, having both ICC and reliability (at n=130) above the median with respect to the ten current measures.

APPENDIX F

Part II

a. METHODS

We will generate patient-mix-adjusted scores for each hospital for each of the 4 models, and calculate 1-R^2 where R is the correlation of final adjusted score for the base model with one of the three alternative models.. We will do this for linear and top-box coded versions of each measure. We will table these results by measure (for each of 11 measures).

b. RESULTS

The results are presented in Appendix H. For all measures, the adjusted means across the four models are very similar for both linear and top-box scoring. The adjusted means correlations between Model 1 and each of the alternate Models are very high. The informativeness (1-R²) is less than 1%, which indicates very small added information resulting from the added PMA (ERA and/or MHP). The Communication with Doctors measure (linear and top-box scoring) and Coordination of Care (top-box) show the biggest impact. However, on average the addition of ERA and MHP matter the same amount on measures linearly scored, while MHP matters a little more than ERA on top-box measures.

c. CONCLUSIONS

Overall, ERA and MHP have very little impact on informativeness. Thus, we recommend against addition of either item as a patient-mix adjuster. However, CMS will continue collection of these variables for analysis and oversight purposes.

APPENDIX G

Part III

a. METHODS

We will produce difference in hospital-level scores under standard PMA model and each of the alternative PMA models. We will show histograms and univariate descriptive statistics of these differences, and list the five hospitals with the biggest gains and losses for each measure, along with mean patient-mix-adjusters for those hospitals.

b. RESULTS

The results are presented in Appendix I. The alternate models have little impact on hospital-level adjusted means. In general, hospitals are effected by less than 1 point with the exception of top-box score Coordination of Care with 1 point on a 0-100 scale.

c. CONCLUSIONS

Overall, ERA and PMA have little effect on hospital scores. Furthermore, there are concerns about the validity of ERA because it is believed to be over-reported by patients, and MHP is moderately correlated with GHP and varies little from hospital to hosptial. For these reasons, we recommend against addition of ERA or MHP as PMA.

APPENDIX H

2. Mode Effects

Part I

1. METHODS

We will add patient-mix adjusters to the two base models that predict Coordination of Care outcome (in linear mean and top-box form) from hospital indicator and the three assigned mode indicators (with Mail Only mode as the reference mode). The top-box version of these patient-mix adjusted mode estimates would potentially be used for adjustment.

b. RESULTS

The CATI (phone) mode mode effects are larger for the CoC measure, with and without PMA. These mode effects are larger for top-box scored measures, ranging from 3.13 to 3.50 for with and without PMA, respectively. In contrast, the effects for IVR and Mixed mode are small.

c. CONCLUSIONS

Findings are similar to Mode Experiment I, with CATI having a larger adjustment for the new measure Coordination of Care..

APPENDIX I

METHODS

To evaluate heterogeneity of survey mode affects within hospitals, we will run mixed effect models with mode and standard PMA as fixed effects, hospital and hospital by mode random effects.

RESULTS

Since the interactions of CATI by hospital and Mixed mode by hospital are not significant there is no evidence that modes CATI and Mixed vary much from hospital to hospital with respect to CoC. However, the IVR variance component is significant indicating some variability in IVR mode from hospital to hospital.

Appendices

Appendix A.1

		N					Mean					SD
		Overall	CATI	Mail	IVR	Mixed	Overall	CATI	Mail	IVR	Mixed	Overall	CATI	Mail	IVR	Mixed
q23	Staff took preferences into account in deciding what care needs would be when I left	6098	1661	1478	1060	1899	3.27	3.32	3.25	3.19	3.29	0.71	0.65	0.71	0.78	0.70
q24	When I left the hospital, had good understanding of things I was responsible for managing health	6256	1733	1501	1064	1958	3.41	3.47	3.37	3.35	3.40	0.67	0.62	0.66	0.75	0.67
q25	When I left the hospital, clearly understood the purpose for taking each of my medications	5233	1357	1308	941	1627	3.49	3.54	3.46	3.44	3.49	0.66	0.58	0.67	0.72	0.67
care	Coordination of Care Composite (3)	6304	1742	1510	1071	1981	79.19	80.85	78.45	77.19	79.37	19.78	17.67	20.15	22.11	19.80
care_tb	Coordination of Care top-box (average of top-box items)	6304	1742	1510	1071	1981	47.13	48.82	45.31	45.86	47.72	42.45	41.48	43.22	42.81	42.48

Appendix A.2

Table of Q23 by MODE

Q23(Staff took preferences into account in deciding what care needs would be when I left)

MODE(Survey Mode (CATI/Telephone only, Mail, MIXED, TT-IVR)

Frequency ‚

Col Pct ‚CATI ‚Mail ‚TT-IVR ‚MIXED ‚ Total

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

. ‚ 90 ‚ 87 ‚ 52 ‚ 125 ‚ 354

‚ 5.14 ‚ 5.56 ‚ 4.68 ‚ 6.18 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

1 Strongly ‚ 29 ‚ 44 ‚ 49 ‚ 53 ‚ 175

disagree ‚ 1.66 ‚ 2.81 ‚ 4.41 ‚ 2.62 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

2 Disagree ‚ 88 ‚ 105 ‚ 96 ‚ 113 ‚ 402

‚ 5.03 ‚ 6.71 ‚ 8.63 ‚ 5.58 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

3 Agree ‚ 872 ‚ 773 ‚ 516 ‚ 964 ‚ 3125

‚ 49.80 ‚ 49.39 ‚ 46.40 ‚ 47.63 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

4 Strongly agree ‚ 672 ‚ 556 ‚ 399 ‚ 769 ‚ 2396

‚ 38.38 ‚ 35.53 ‚ 35.88 ‚ 37.99 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

Total 1751 1565 1112 2024 6452

Table of Q24 by MODE

Q24(When I left the hospital, had good understanding of things I was responsible for

managing health)

MODE(Survey Mode (CATI/Telephone only, Mail, MIXED, TT-IVR)

Frequency ‚

Col Pct ‚CATI ‚Mail ‚TT-IVR ‚MIXED ‚ Total

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

. ‚ 18 ‚ 64 ‚ 48 ‚ 66 ‚ 196

‚ 1.03 ‚ 4.09 ‚ 4.32 ‚ 3.26 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

1 Strongly ‚ 22 ‚ 25 ‚ 38 ‚ 43 ‚ 128

disagree ‚ 1.26 ‚ 1.60 ‚ 3.42 ‚ 2.12 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

2 Disagree ‚ 46 ‚ 81 ‚ 61 ‚ 75 ‚ 263

‚ 2.63 ‚ 5.18 ‚ 5.49 ‚ 3.71 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

3 Agree ‚ 764 ‚ 706 ‚ 452 ‚ 890 ‚ 2812

‚ 43.63 ‚ 45.11 ‚ 40.65 ‚ 43.97 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

4 Strongly agree ‚ 901 ‚ 689 ‚ 513 ‚ 950 ‚ 3053

‚ 51.46 ‚ 44.03 ‚ 46.13 ‚ 46.94 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

Total 1751 1565 1112 2024 6452

Table of Q25 by MODE

Q25(When I left the hospital, clearly understood the purpose for taking each of my

medications)

MODE(Survey Mode (CATI/Telephone only, Mail, MIXED, TT-IVR)

Frequency ‚

Col Pct ‚CATI ‚Mail ‚TT-IVR ‚MIXED ‚ Total

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

. ‚ 38 ‚ 65 ‚ 58 ‚ 64 ‚ 225

‚ 2.17 ‚ 4.15 ‚ 5.22 ‚ 3.16 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

Not given meds ‚ 356 ‚ 192 ‚ 113 ‚ 333 ‚ 994

‚ 20.33 ‚ 12.27 ‚ 10.16 ‚ 16.45 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

1 Strongly ‚ 14 ‚ 26 ‚ 31 ‚ 39 ‚ 110

disagree ‚ 0.80 ‚ 1.66 ‚ 2.79 ‚ 1.93 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

2 Disagree ‚ 19 ‚ 56 ‚ 37 ‚ 46 ‚ 158

‚ 1.09 ‚ 3.58 ‚ 3.33 ‚ 2.27 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

3 Agree ‚ 540 ‚ 513 ‚ 364 ‚ 627 ‚ 2044

‚ 30.84 ‚ 32.78 ‚ 32.73 ‚ 30.98 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

4 Strongly agree ‚ 784 ‚ 713 ‚ 509 ‚ 915 ‚ 2921

‚ 44.77 ‚ 45.56 ‚ 45.77 ‚ 45.21 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

Total 1751 1565 1112 2024 6452

Appendix A.3

Table of Q23 by MODE

Q23(Staff took preferences into account in deciding what care needs would be when I left)

MODE(Survey Mode (CATI/Telephone only, Mail, MIXED, TT-IVR)

Frequency ‚

Col Pct ‚CATI ‚Mail ‚TT-IVR ‚MIXED ‚ Total

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

1 Strongly ‚ 29 ‚ 44 ‚ 49 ‚ 53 ‚ 175

disagree ‚ 1.75 ‚ 2.98 ‚ 4.62 ‚ 2.79 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

2 Disagree ‚ 88 ‚ 105 ‚ 96 ‚ 113 ‚ 402

‚ 5.30 ‚ 7.10 ‚ 9.06 ‚ 5.95 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

3 Agree ‚ 872 ‚ 773 ‚ 516 ‚ 964 ‚ 3125

‚ 52.50 ‚ 52.30 ‚ 48.68 ‚ 50.76 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

4 Strongly agree ‚ 672 ‚ 556 ‚ 399 ‚ 769 ‚ 2396

‚ 40.46 ‚ 37.62 ‚ 37.64 ‚ 40.49 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

Total 1661 1478 1060 1899 6098

Frequency Missing = 354

Table of Q24 by MODE

Q24(When I left the hospital, had good understanding of things I was responsible for

managing health)

MODE(Survey Mode (CATI/Telephone only, Mail, MIXED, TT-IVR)

Frequency ‚

Col Pct ‚CATI ‚Mail ‚TT-IVR ‚MIXED ‚ Total

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

1 Strongly ‚ 22 ‚ 25 ‚ 38 ‚ 43 ‚ 128

disagree ‚ 1.27 ‚ 1.67 ‚ 3.57 ‚ 2.20 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

2 Disagree ‚ 46 ‚ 81 ‚ 61 ‚ 75 ‚ 263

‚ 2.65 ‚ 5.40 ‚ 5.73 ‚ 3.83 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

3 Agree ‚ 764 ‚ 706 ‚ 452 ‚ 890 ‚ 2812

‚ 44.09 ‚ 47.04 ‚ 42.48 ‚ 45.45 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

4 Strongly agree ‚ 901 ‚ 689 ‚ 513 ‚ 950 ‚ 3053

‚ 51.99 ‚ 45.90 ‚ 48.21 ‚ 48.52 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

Total 1733 1501 1064 1958 6256

Frequency Missing = 196

Table of Q25 by MODE

Q25(When I left the hospital, clearly understood the purpose for taking each of my

medications)

MODE(Survey Mode (CATI/Telephone only, Mail, MIXED, TT-IVR)

Frequency ‚

Col Pct ‚CATI ‚Mail ‚TT-IVR ‚MIXED ‚ Total

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

1 Strongly ‚ 14 ‚ 26 ‚ 31 ‚ 39 ‚ 110

disagree ‚ 1.03 ‚ 1.99 ‚ 3.29 ‚ 2.40 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

2 Disagree ‚ 19 ‚ 56 ‚ 37 ‚ 46 ‚ 158

‚ 1.40 ‚ 4.28 ‚ 3.93 ‚ 2.83 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

3 Agree ‚ 540 ‚ 513 ‚ 364 ‚ 627 ‚ 2044

‚ 39.79 ‚ 39.22 ‚ 38.68 ‚ 38.54 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

4 Strongly agree ‚ 784 ‚ 713 ‚ 509 ‚ 915 ‚ 2921

‚ 57.77 ‚ 54.51 ‚ 54.09 ‚ 56.24 ‚

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒˆ

Total 1357 1308 941 1627 5233

Frequency Missing = 1219

Appendix B.1

Linear coding

Cronbach Coefficient Alpha

Variables Alpha

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

Raw 0.818870

Standardized 0.819706

Pearson Correlation Coefficients, N = 5064

Prob > |r| under H0: Rho=0

Q23 Q24 Q25

Q23 1.00000 0.64079 0.53106

Staff took preferences into account in deciding <.0001 <.0001

what care needs would be when I left

Q24 0.64079 1.00000 0.63554

When I left the hospital, had good understanding <.0001 <.0001

of things I was responsible for managing health

Q25 0.53106 0.63554 1.00000

When I left the hospital, clearly understood <.0001 <.0001

the purpose for taking each of my medications

Top-box coding

Cronbach Coefficient Alpha

Variables Alpha

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

Raw 0.804103

Standardized 0.803886

Pearson Correlation Coefficients, N = 5064

Prob > |r| under H0: Rho=0

rq23_tb rq24_tb rq25_tb

rq23_tb 1.00000 0.60453 0.50339

(Top-box) Staff took preferences into account in <.0001 <.0001

deciding what care needs would be when I left

rq24_tb 0.60453 1.00000 0.62431

(Top-box) When I left the hospital, had good understanding <.0001 <.0001

of things I was responsible for managing health

rq25_tb 0.50339 0.62431 1.00000

(Top-box) When I left the hospital, clearly understood <.0001 <.0001

the purpose for taking each of my medications

Appendix B.2

Linear coding

Pearson Correlations:

CoC Q23 Q24 Q25

0.49586 0.45955 0.43899 0.39580

Nurse communication Composite <.0001 <.0001 <.0001 <.0001

6292 6088 6246 5223

0.46687 0.40369 0.43257 0.37435

Doctor communication Composite <.0001 <.0001 <.0001 <.0001

6282 6078 6235 5217

0.39208 0.37004 0.35160 0.29872

Responsiveness of hospital staff Composite <.0001 <.0001 <.0001 <.0001

5742 5570 5698 4840

0.43102 0.40108 0.38750 0.33000

Pain management Composite <.0001 <.0001 <.0001 <.0001

4554 4433 4526 3909

0.49176 0.42734 0.42095 0.44382

Communication about medicines Composite <.0001 <.0001 <.0001 <.0001

3572 3474 3544 3142

0.41069 0.38573 0.36170 0.29870

Discharge information Composite <.0001 <.0001 <.0001 <.0001

5904 5719 5864 4936

0.30425 0.26776 0.27647 0.24951

Quiet <.0001 <.0001 <.0001 <.0001

6263 6062 6217 5199

0.32255 0.32193 0.28734 0.22954

Clean <.0001 <.0001 <.0001 <.0001

6261 6060 6213 5199

0.48049 0.44864 0.42519 0.36346

Rate hospital <.0001 <.0001 <.0001 <.0001

6233 6035 6187 5174

0.50586 0.47413 0.44763 0.38463

Recommend hospital <.0001 <.0001 <.0001 <.0001

6253 6054 6207 5192

Appendix B.3

Top-box coding

Spearman Correlations:

CARE_tb rq23_tb rq24_tb rq25_tb

0.39736 0.34257 0.36234 0.32614

Nurse communication top-box (average of top-box items) <.0001 <.0001 <.0001 <.0001

6292 6088 6246 5223

0.39455 0.32021 0.36502 0.32495

Doctor communication top-box (average of top-box items) <.0001 <.0001 <.0001 <.0001

6282 6078 6235 5217

0.31590 0.27899 0.28785 0.24704

Responsiveness of hospital staff top-box (average of top-box items) <.0001 <.0001 <.0001 <.0001

5742 5570 5698 4840

0.34275 0.28059 0.32299 0.28451

Pain management top-box (average of top-box items) <.0001 <.0001 <.0001 <.0001

4554 4433 4526 3909

0.43816 0.36459 0.37405 0.40300

Communication about medicines top-box (average of top-box items) <.0001 <.0001 <.0001 <.0001

3572 3474 3544 3142

0.30304 0.26795 0.26319 0.23492

Discharge information top-box (average of top-box items) <.0001 <.0001 <.0001 <.0001

5904 5719 5864 4936

0.24378 0.20341 0.22498 0.20806

Quiet top-box (Always) <.0001 <.0001 <.0001 <.0001

6263 6062 6217 5199

0.26302 0.24238 0.23428 0.20292

Clean top-box (Always) <.0001 <.0001 <.0001 <.0001

6261 6060 6213 5199

0.39271 0.33693 0.35120 0.32085

Rate hospital top-box (9 or 10) <.0001 <.0001 <.0001 <.0001

6233 6035 6187 5174

0.40812 0.34280 0.37075 0.33439

Recommend hospital top-box (Definitely Yes) <.0001 <.0001 <.0001 <.0001

6253 6054 6207 5192

Appendix C.1 (linear measures)

	Coordination of Care Composite
	Base Model	Model 1	Model 2	Model 3
ER admit		-1.58 **		-1.49 *
MHP			-2.33 ***	-2.32 ***
Maternity	9.82 ***	8.78 ***	9.39 ***	8.41 **
Surgical	6.66 **	6.08 **	6.67 **	6.12 **
Age:
18-24	0.16	0.13	0.11	0.08
25-34	2.28	2.01	2.34	2.09
35-44	2.29	2.11	2.5	2.32
45-54	2.38	2.31	2.38	2.32
55-64	3.31 **	3.21 **	3.27 **	3.18 **
65-74	3.93 ***	3.84 ***	3.72 ***	3.64 ***
75-84	1.42	1.38	1.36	1.32
Education	-0.02	-0.05	-0.28	-0.31
GHP	-2.71 ***	-2.67 ***	-1.62 ***	-1.58 ***
Language other than English	-1.08	-0.98	-1.18	-1.09
Response Percentile	-1.23	-1.17	-1.16	-1.11
Maternity*Age	-2.43 *	-2.3 *	-2.42 *	-2.3 *
Surgical*Age	-0.77 *	-0.8 *	-0.82 *	-0.85 *

*** p<0.001, ** p<0.01 and * p<0.

Appendix C.2 (top-box measures)

	Coordination of Care Composite
	Base Model	Model 1	Model 2	Model 3
ER admit		-2.08		-1.86
MHP			-5.89 ***	-5.88 ***
Maternity	18.43 ***	17.07 **	17.33 **	16.12 **
Surgical	8.35	7.59	8.38	7.7
Age:
18-24	5.77	5.73	5.64	5.6
25-34	9.56 **	9.2 **	9.72 **	9.4 **
35-44	10.4 **	10.16 **	10.91 ***	10.7 ***
45-54	11.97 ***	11.88 ***	11.98 ***	11.9 ***
55-64	12.7 ***	12.57 ***	12.6 ***	12.49 ***
65-74	10.92 ***	10.8 ***	10.39 ***	10.29 ***
75-84	3.19	3.13	3.04	2.99
Education	0.89 *	0.84	0.23	0.19
GHP	-5.55 ***	-5.49 ***	-2.78 ***	-2.73 ***
Language other than English	-4.82	-4.68	-5.06	-4.94
Response Percentile	-4.02 *	-3.95 *	-3.84 *	-3.78 *
Maternity*Age	-5.31 *	-5.14 *	-5.29 *	-5.14 *
Surgical*Age	-0.68	-0.72	-0.8	-0.83

*** p<0.001, ** p<0.01 and * p<0.05

Appendix D

Mode (Mail only as reference):	CARE model1	CARE model2	CARE model3	CARE model4	CARE tb model1	CARE tb model2	CARE tb model3	CARE tb model4
CATI	2.14 **	2.17 **	2.07 **	2.1 **	3.33 *	3.37 *	3.16 *	3.19 *
MIXED	1	1	0.97	0.97	2.96 *	2.97 *	2.89 *	2.89 *
IVR	-2.03 **	-2 **	-2.28 **	-2.24 **	-1.05	-1	-1.65	-1.6

*** p<0.001, ** p<0.01 and * p<0.

(This Table is based on data with all correct exclusions)

Appendix E

Hospital-level InterClass Correlation (ICC), Spearman Brown Correlation (n=300) and Reliability

Appendix F.1

Patient-Mix Adjusted Means Scores for Linearly Scored Measures

PROVID	CARE_1	CARE_2	CARE_3	CARE_4
010024	80.01	80.02	80.05	80.05
030037	75.36	75.24	75.38	75.27
030122	79.82	79.63	79.87	79.69
040014	78.34	78.29	78.37	78.32
040027	79.71	79.61	79.98	79.89
050069	79.28	79.29	79.46	79.47
050104	78.63	78.7	78.87	78.93
050438	84.81	84.72	84.72	84.64
050455	78.56	78.51	78.65	78.61
050506	80	79.93	80.16	80.1
050746	74.93	75.06	74.94	75.06
060030	83.94	83.86	83.89	83.82
100113	81.67	81.52	81.78	81.65
100135	79.9	79.79	80.04	79.94
100187	65.07	65.06	64.87	64.86
100189	76.46	76.69	76.53	76.75
110075	78.88	78.89	78.72	78.73
110198	71.71	71.9	71.69	71.87
120006	85.18	85.46	85.55	85.81
150012	81.24	81.17	81.31	81.25
150150	80.45	80.15	80.47	80.18
190046	79.31	79.3	79.32	79.3
210019	79.69	79.62	79.41	79.35
210040	76.71	76.83	76.92	77.03
230002	78.68	78.7	78.52	78.54
230070	84.43	84.5	84.43	84.49
230236	82.17	82.29	81.92	82.03
240053	79.11	79.02	79.12	79.04
260027	76.23	76.11	76.54	76.43
260094	78.16	78.43	78.19	78.45
290022	68.45	68.71	68.39	68.63
290041	75.07	75.1	75.02	75.05
310081	79.42	79.53	79.08	79.19
360012	81.63	81.6	81.63	81.59
360155	81.8	81.82	81.65	81.67
370008	77.49	77.55	77.51	77.57
370149	81.52	81.66	81.86	81.99
390211	79.47	79.47	79.17	79.18
440034	80.53	80.58	80.64	80.69
440091	83.15	82.96	83.13	82.95
450056	80.47	80.42	80.56	80.51
450424	78.96	79.09	78.58	78.71
450647	80.46	80.35	80.52	80.42
450675	77.99	78.06	78.21	78.28
450697	73.43	73.42	72.89	72.89
450742	76.27	76.27	76.35	76.34
490075	77.4	77.4	77.22	77.23

Note: Model 1 is composed of the standard patient-mix adjustor; Model 2 adds to Model 1 the indicator for ER admission; Model 3 adds to Model 1 linear MHP; Model 4 adds to Model 1 the indicator for ER admission and linear MHP.

Patient-Mix Adjusted Means Scores for Top-Box Scored Measures

PROVID	CARE_tb_1	CARE_tb_2	CARE_tb_3	CARE_tb_4
010024	46.38	46.39	46.47	46.48
030037	37.52	37.36	37.56	37.43
030122	50.78	50.54	50.92	50.7
040014	47.53	47.47	47.61	47.55
040027	47.33	47.2	48.02	47.9
050069	48.84	48.86	49.3	49.31
050104	43.2	43.29	43.8	43.88
050438	57.78	57.67	57.56	57.46
050455	46.46	46.4	46.71	46.66
050506	47.79	47.7	48.21	48.13
050746	40.95	41.12	40.97	41.12
060030	55.47	55.37	55.35	55.26
100113	51.85	51.67	52.15	51.98
100135	48.75	48.62	49.12	49
100187	37.87	37.86	37.36	37.35
100189	40.32	40.61	40.5	40.76
110075	44.69	44.7	44.29	44.29
110198	29.14	29.38	29.08	29.29
120006	57.96	58.33	58.9	59.22
150012	51.63	51.54	51.82	51.74
150150	47.98	47.59	48.03	47.68
190046	49.57	49.55	49.58	49.56
210019	47.46	47.38	46.75	46.68
210040	42.02	42.18	42.55	42.69
230002	44.03	44.06	43.63	43.66
230070	57.91	58	57.91	57.99
230236	54.27	54.43	53.63	53.77
240053	45.32	45.21	45.35	45.25
260027	40.74	40.59	41.52	41.38
260094	46.97	47.32	47.04	47.36
290022	30.67	31	30.52	30.81
290041	40.43	40.48	40.32	40.36
310081	47.42	47.57	46.57	46.7
360012	52.63	52.59	52.62	52.58
360155	54.7	54.72	54.32	54.34
370008	41.45	41.53	41.51	41.59
370149	50.89	51.07	51.73	51.89
390211	44.29	44.3	43.54	43.55
440034	51.02	51.09	51.31	51.38
440091	53.87	53.63	53.82	53.6
450056	49.45	49.38	49.66	49.6
450424	45.87	46.03	44.91	45.06
450647	48.61	48.47	48.76	48.63
450675	44.64	44.73	45.21	45.29
450697	37.17	37.15	35.8	35.79
450742	41.82	41.81	42.01	41.99
490075	43.97	43.98	43.53	43.54

Note: Model 1 is composed of the standard patient-mix adjustor; Model 2 adds to Model 1 the indicator for ER admission; Model 3 adds to Model 1 linear MHP; Model 4 adds to Model 1 the indicator for ER admission and linear MHP.

Appendix F.2

Correlation and Informativeness (1 - R²) of Adjusted Means from Model 1 with Each of the Three Alternative Models.

	Correlation			1 - R²
	Model 1 & 2	Model 1 & 3	Model 1 & 4	Model 1 & 2	Model 1 & 3	Model 1 & 4
Linear
Nurse communication Composite	1.000	0.999	0.999	0.000	0.001	0.001
Doctor communication Composite	0.997	0.998	0.995	0.007	0.003	0.010
Responsiveness of hospital staff Composite	1.000	1.000	1.000	0.000	0.000	0.001
Pain management Composite	1.000	0.999	0.999	0.000	0.002	0.002
Communication about medicines Composite	1.000	0.999	0.999	0.000	0.002	0.002
Discharge information Composite	0.999	1.000	0.999	0.002	0.000	0.002
Quiet	1.000	1.000	1.000	0.000	0.000	0.001
Clean	1.000	1.000	1.000	0.000	0.000	0.000
Rate hospital	1.000	1.000	1.000	0.000	0.000	0.001
Recommend hospital	1.000	1.000	1.000	0.000	0.000	0.000
Coordination of Care Composite	0.999	0.999	0.998	0.001	0.002	0.003
Top-Box
Nurse communication	1.000	0.999	0.999	0.000	0.003	0.003
Doctor communication Composite	0.996	0.998	0.995	0.007	0.004	0.010
Responsiveness of hospital staff Composite	1.000	0.999	0.999	0.000	0.001	0.001
Pain management Composite	1.000	0.999	0.999	0.000	0.002	0.003
Communication about medicines Composite	1.000	0.998	0.999	0.000	0.003	0.003
Discharge information Composite	0.999	1.000	0.999	0.002	0.000	0.002
Quiet	1.000	1.000	1.000	0.000	0.001	0.001
Clean	1.000	1.000	1.000	0.000	0.001	0.001
Rate hospital	1.000	1.000	0.999	0.000	0.001	0.001
Recommend hospital	1.000	1.000	1.000	0.000	0.001	0.001
Coordination of Care Composite	1.000	0.997	0.997	0.001	0.005	0.006

Note: Model 1 is composed of the standard patient-mix adjustor; Model 2 adds to Model 1 the indicator for ER admission; Model 3 adds to Model 1 linear MHP; Model 4 adds to Model 1 the indicator for ER admission and linear MHP.

Appendix G

Difference in adjusted mean scores from standard PMA Model (Base Model) and each of the alternate Models 2, 3 and 4.

DUPONT HOSPITAL LLC (Gain: -0.299)	SOUTHWEST GENERAL HOSPITAL (Gain: -0.539)	SOUTHWEST GENERAL HOSPITAL (Gain: -0.547)
BANNER GATEWAY MEDICAL CENTER (Gain: -0.19)	SAN JACINTO METHODIST HOSPITAL (Gain: -0.379)	PENINSULA REGIONAL MEDICAL CENTER (Gain: -0.339)
MEMORIAL HOSPITAL, CHATTANOOGA, TN (Gain: -0.187)	UNDERWOOD - MEMORIAL HOSPITAL (Gain: -0.336)	SHARON REGIONAL HEALTH SYSTEM (Gain: -0.29)
SHANDS HOSPITAL AT THE UNIVERSITY OF FLORIDA (Gain: -0.143)	SHARON REGIONAL HEALTH SYSTEM (Gain: -0.299)	DUPONT HOSPITAL LLC (Gain: -0.262)
RESEARCH MEDICAL CENTER (Gain: -0.119)	PENINSULA REGIONAL MEDICAL CENTER (Gain: -0.281)	SAN JACINTO METHODIST HOSPITAL (Gain: -0.257)
NORTH FULTON REGIONAL HOSPITAL (Loss: 0.185)	SAINT FRANCIS MEDICAL CENTER (Loss: 0.237)	MEDICAL CENTER OF ARLINGTON (Loss: 0.29)
NORTHWEST MEDICAL CENTER (Loss: 0.227)	BAXTER REGIONAL MEDICAL CENTER (Loss: 0.274)	SAINT FRANCIS MEDICAL CENTER (Loss: 0.297)
DESERT SPRINGS HOSPITAL CENTER (Loss: 0.254)	RESEARCH MEDICAL CENTER (Loss: 0.307)	NORTHWEST HOSPITAL CENTER (Loss: 0.325)
SKAGGS COMMUNITY HEALTH CENTER (Loss: 0.272)	UNITY HEALTH CENTER (Loss: 0.336)	UNITY HEALTH CENTER (Loss: 0.466)
CASTLE MEDICAL CENTER (Loss: 0.28)	CASTLE MEDICAL CENTER (Loss: 0.371)	CASTLE MEDICAL CENTER (Loss: 0.634)

DUPONT HOSPITAL LLC (Gain: -0.387)	SOUTHWEST GENERAL HOSPITAL (Gain: -1.361)	SOUTHWEST GENERAL HOSPITAL (Gain: -1.371)
BANNER GATEWAY MEDICAL CENTER (Gain: -0.246)	SAN JACINTO METHODIST HOSPITAL (Gain: -0.957)	SAN JACINTO METHODIST HOSPITAL (Gain: -0.808)
MEMORIAL HOSPITAL, CHATTANOOGA, TN (Gain: -0.243)	UNDERWOOD - MEMORIAL HOSPITAL (Gain: -0.849)	PENINSULA REGIONAL MEDICAL CENTER (Gain: -0.78)
SHANDS HOSPITAL AT THE UNIVERSITY OF FLORIDA (Gain: -0.186)	SHARON REGIONAL HEALTH SYSTEM (Gain: -0.755)	SHARON REGIONAL HEALTH SYSTEM (Gain: -0.744)
RESEARCH MEDICAL CENTER (Gain: -0.154)	PENINSULA REGIONAL MEDICAL CENTER (Gain: -0.709)	UNDERWOOD - MEMORIAL HOSPITAL (Gain: -0.715)
NORTH FULTON REGIONAL HOSPITAL (Loss: 0.24)	SAINT FRANCIS MEDICAL CENTER (Loss: 0.599)	MEDICAL CENTER OF ARLINGTON (Loss: 0.649)
NORTHWEST MEDICAL CENTER (Loss: 0.294)	BAXTER REGIONAL MEDICAL CENTER (Loss: 0.692)	NORTHWEST HOSPITAL CENTER (Loss: 0.671)
DESERT SPRINGS HOSPITAL CENTER (Loss: 0.329)	RESEARCH MEDICAL CENTER (Loss: 0.776)	SAINT FRANCIS MEDICAL CENTER (Loss: 0.672)
SKAGGS COMMUNITY HEALTH CENTER (Loss: 0.352)	UNITY HEALTH CENTER (Loss: 0.848)	UNITY HEALTH CENTER (Loss: 1.007)
CASTLE MEDICAL CENTER (Loss: 0.364)	CASTLE MEDICAL CENTER (Loss: 0.937)	CASTLE MEDICAL CENTER (Loss: 1.259)

Appendix H

Mode Effects with and without PMA adjustment.

		LINEAR SCORING			TOP-BOX SCORING
		CATI	IVR	MIXED	CATI	IVR	MIXED
CARE	Without PMA	2.25 **	-1.53	0.89	3.50 *	0.12	2.53
	With PMA	2.10 **	-1.97 *	0.88	3.13 *	-0.77	2.45

Appendix I

Mixed effect models with mode and standard PMA as fixed effects, and hospital and hospital by mode as random effects.

	CARE
RANDOM EFFECTS:	SE	VAR
PROVID	2.65	7.01	**
CATI*PROVID	0.00	0.00
MIXED*PROVID	0.60	0.36
IVR*PROVID	4.07	16.54	*
Residual	19.06	363.4	***

1 The PMA are indicators for age (18-24, 25-34, 35-44, 45-55, 56-64, 65-74, 75-84 and 85 or older (reference)), linear education, linear overall health, indicators for maternity and surgical service lines, indicator for language other than English spoken at home, response percentile, and interaction of linear age with maternity and surgical.