Nancy L. Novotny, PhD(c)

Patient Inclusion Criteria

Trained research coordinators identified all consecutive adult inpatients who were discharged to the patient's home by internal medicine hospitalists. Patient inclusion in our cluster randomized trial required a probability of repeat admission (Pra) score of at least 0.40.11, 12 Hence, the patients in the scale analysis cohort had the same high probability for repeat admission. The research coordinators calculated the Pra within 2 days before discharge from the index hospitalization. The Pra score came from a logistic model of age, sex, prior hospitalizations, prior doctor visits, self‐rated health status, informal caregiver, and comorbid coronary heart disease and diabetes mellitus.11, 12

Patient Exclusion Criteria

We excluded patients if the discharge destination was a nursing home, another acute care hospital, or an inpatient rehabilitation unit. Patients were excluded if life expectancy was less than 6 months as estimated by the hospitalist. Because follow‐up occurred via interview, patients without telephones or English‐ or Spanish‐language skills were excluded. Patients with cognitive impairment could participate with consent from a legally authorized representative and with a proxy whospent a minimum of 3 hours daily with the patient and was willing to answer postdischarge interviews.

Baseline Assessment

During the index hospital admission, trained data abstractors recorded baseline patient data: age, sex, race, diabetes mellitus, heart failure, chronic obstructive pulmonary disease, and coronary heart disease. Patients or proxies provided the number of hospital admissions and doctor visits during the year before the index hospital admission. We recorded the availability of an informal caregiver in response to the question Is there a friend, relative, or neighbor who would take care of you for a few days if necessary? Patients rated their health status on the following scale: poor, fair, good, very good, or excellent.

Discharge and Postdischarge Procedures

At the end of the index hospitalization, hospitalists and ward nurses used standardized forms for discharge diagnoses, prescriptions, instructions, and appointments. Discharge‐planning nurses or social workers consulted with hospitalists and ward nurses and then coordinated service providers including home health nurses, physical therapists, home health aides, homemaker service providers, durable medical equipment vendors, home oxygen vendors, home infusion pharmacists, social workers, rehabilitation service providers, legal aide providers, and others. After discharge, trained research personnel conducted 2 telephone interviews with the patient or the patient's proxy. The first interview occurred 1 week after discharge. Interviewers read verbatim items from the PREPARED10 and the Satisfaction with Information about Medicines Scale (SIMS).13 During the second telephone interview 30 10 days after discharge, interviewers recorded if patients had experienced at least 1 emergency department visit during the month after discharge.

The purpose of the PREPARED items was to have a bank of items and responses that could be used to generate the B‐PREPARED scale. The PREPARED questionnaire was originally developed to provide feedback to hospital ward staff about the quality of discharge‐planning activities that occurred during hospitalization.9 Discriminant factor analysis on the original 16 process questions revealed 4 factors that explained 57% of the total variance in patient/caregiver responses. The PREPARED domains included information exchange on community services and equipment, management of medication, the process of preparing to cope after discharge, and having control over one's discharge circumstances.9 The purpose of the SIMS was as a construct to compare with the B‐PREPARED scale. The derivation and validity of the SIMS have been described extensively elsewhere.13 In summary, the SIMS items were derived from recommendations of the Association of the British Pharmaceutical Industry. The intent of the SIMS was to determine if a patient's medication information needs were met and to allow comparison between patients or groups. Respondents selected 1 of 5 options for each of the 17 items. The sum of scores for each of the SIMS items yielded a total score that ranged from 0 to 17. Patients with high total SIMS scores had high satisfaction with the amount of medication information they received. Validation samples included inpatients and outpatients with a variety of diseases and characteristics. SIMS demonstrated adequate internal consistency, test‐retest reliability, and criterion‐related validity.13

Item Selection and Scoring of the B‐PREPARED Instrument

We selected an initial pool of items from the PREPARED instrument.10 The goal was a parsimonious, comprehensive, and valid instrument for use in clinical and research environments. When we retained or deleted items, our decision process was conservative, conceptual, and statistical. We performed item reduction in the following steps defined a priori. First, we agreed on items consistent with domains in the prepared for discharge construct as defined by expert consensus.9 Second, we excluded items that assessed qualities of the discharge process that were imperceptible to the patient on the day of discharge. Third, we excluded items that elicited open‐ended responses unsuitable for quantitative scale development and analysis. Fourth, we assessed reliability as defined by the Cronbach's alpha statistic. We excluded items that substantially decreased Cronbach's alpha.

Measures of Construct Validity

We used 2 measures of construct validity in our assessment of B‐PREPARED. One construct was patient worry. During the interview 1 week after discharge, research personnel asked, Now that you have been out of the hospital for a while, has anything been worrying you about managing at home? Response options for the dichotomous worry item were no or yes. We anticipated worried patients would have lower B‐PREPARED scale values. The other construct, SIMS, evaluated patient preparedness related to medication information exchange. The hypothesis was a positive correlation between SIMS and B‐PREPARED scale values.

Measure of Predictive Validity

We asked if B‐PREPARED predicted and discriminated groups of patients who did or did not visit emergency departments after hospital discharge. Emergency department visits were relevant adverse outcomes because of their association with postdischarge adverse events due to inpatient treatment.4 Emergency department visits reflected new or worsening symptoms after discharge. In our scale analysis, the hypothesis was patients with at least 1 emergency department visit would have lower B‐PREPARED scale values.

Analysis

Analyses were performed with SPSS PC (Version 14.0.2, SPSS Inc, Chicago, IL). We reported descriptive statistics as means, standard deviations, and range for interval variables and percentages for nominal variables. To determine the internal consistency of the scale, we calculated Cronbach's alpha. We assessed the distribution of the B‐PREPARED scale with visual and statistical tests for skewness. While using the SPSS FACTOR program, we performed principal components extractions and then rotated components using the oblique promax technique. Component scores were saved using the regression score procedure. Component loadings above 0.30 were considered important. Statistical inference tests were the Mann‐Whitney U for median differences between 2 groups and the Spearman correlation for associations. We reported medians with 25th and 75th percentiles. Differences between 2 correlations were tested using Fischer z transformations. The accepted level of significance was P < 0.05.

Description of Cohort

We approached 5124 patients during the index hospital admission. After applying exclusion criteria, we obtained consent and enrolled 491 patients. The reasons for exclusion were low Pra score for 34.9% of ineligible patients, discharge to nursing home for 12.8%, declined consent for 10.8%, nonparticipating hospitalist service for 9.1%, discharged during screen for 8.5%, previously enrolled in study for 5.6%, and declined screening for 2.3%. Each of the other exclusion criteria accounted for less than 4% of the ineligible patients. After subtracting 6% of eligible patients (31 of 491) who died, withdrew, or were lost during the first month, there were 460 patients available for analysis. Table 1 describes the patients' characteristics. Most of the patients, 75.2% (346 of 460), were less than 65 years old, and the mean age was 53.9 15.5 years. Many patients had chronic diseases including diabetes mellitus, coronary heart disease, heart failure, and chronic obstructive pulmonary disease. Most patients, 81.5% (375 of 460), rated their health as poor or fair, and 53.5% (246/460) had 1 or more hospital admissions during the year before their index admission. Cohort patients had a high probability of repeat admission: mean Pra 0.49 0.07 (range 0.400.70).

Item Reduction, Internal Consistency, and Score Distributions

Item reduction resulted in 12 items that fulfilled conceptual criteria. Table 2 shows the items and the distribution of responses. One of the 12 items, delays on the day you left the hospital (item 12, Table 2), was deleted because the item depressed the Cronbach's alpha. The B‐PREPARED with 11 items had acceptable internal consistency for the full cohort (Cronbach's alpha = 0.76).

For an individual patient, the sum of the scores for each item yielded a B‐PREPARED scale value. In the 460‐patient cohort, B‐PREPARED scale values had a mean of 17.3 4.3 and a negatively skewed distribution. A high scale value reflected high perception of discharge preparedness. Each of the 11 items correlated significantly with the B‐PREPARED scale value (P < 0.001, 2‐tailed).

There were substantial ceiling effects with individual items but not in the B‐PREPARED total score. Five of the 9 items with 3 response options had a ceiling effect above 70%. Three items had a dichotomous response option (items 7, 8, and 9). In 2 of these 3 items, more than 90% of respondents selected the response indicating higher preparedness. The total B‐PREPARED did not have noteworthy floor or ceiling effects. In this sample's total B‐PREPARED scores, 0.2% of respondents had the lowest score of 3, and 20% had the highest score of 22.

Principal Component Analysis

In the component analysis, we evaluated the correlation matrix of the 11 items in the B‐PREPARED scale. A Kaiser‐Meyer‐Olkin statistic of 0.76 indicated sufficient sampling adequacy to extract components from the matrix. Principal components extracted 54.2% of the variance associated with the 11‐item B‐PREPARED scale. After inspection of scree plots, we determined that 3 components were extracted before the eigenvalue fell below 1. The pattern matrix for the promax rotation was inspected, and the factor loading of each item appears in Table 3. The item content identified the first component as self‐care information for medications and activities. The second component was equipment and services. The third component was confidence. All B‐PREPARED items loaded primarily on 1 of the 3 components (Table 3).

Construct Validity

We assessed 2 constructs: worry and satisfaction with medication information (SIMS). In the cohort, 25% of patients (115 of 460) reported worry about managing at home. Worried patients had significantly lower B‐PREPARED scale values (median [25%, 75%] = 14 [10, 16]) than patients who did not worry (median [25%, 75%] = 17 [14, 20], P < .001). We calculated SIMS and then correlated SIMS with B‐PREPARED and components. In the cohort, the mean SIMS was 12.1 4.7 (range 0‐17). Patients with greater satisfaction on the SIMS also had higher B‐PREPARED scale values (rho = 0.45, P < 0.001). There was a significant positive correlation between SIMS and the B‐PREPARED component called self‐care information for medications and activities (rho = 0.46, P < 0.001). The other 2 B‐PREPARED components, equipment/services and confidence, were positively correlated with SIMS at much lower levels (rho = 0.18 and rho = 0.24, respectively, both Fischer z transformations P < .001). The B‐PREPARED scale demonstrated validity with the constructs of worry and satisfaction with medication information.

Predictive Validity

We assessed the capacity of the B‐PREPARED to predict and discriminate groups of patients who did or did not visit emergency departments. Within 1 month of hospital discharge, 16.5% of the cohort (76 of 460 patients) had at least 1 visit to an emergency department. B‐PREPARED scale values were lower for those patients who visited emergency departments (median [25%, 75%] = 14 [12, 18]) than those who did not (median [25%, 75%] = 16 [13, 19], P = 0.011). The B‐PREPARED scale analysis supported the hypothesized relationship with emergency department visits.

Correlations between B‐PREPARED and Baseline Characteristics

We evaluated the correlations between a patient's B‐PREPARED scale value and baseline characteristics, shown in Table 1. There was a weak positive correlation with self‐rated health status (rho 0.17, P < .001). Patients who perceived better health status had higher B‐PREPARED scale values than those with poorer status. The other baseline characteristics in Table 1 were not associated with B‐PREPARED scale values.

Patient Inclusion Criteria

Trained research coordinators identified all consecutive adult inpatients who were discharged to the patient's home by internal medicine hospitalists. Patient inclusion in our cluster randomized trial required a probability of repeat admission (Pra) score of at least 0.40.11, 12 Hence, the patients in the scale analysis cohort had the same high probability for repeat admission. The research coordinators calculated the Pra within 2 days before discharge from the index hospitalization. The Pra score came from a logistic model of age, sex, prior hospitalizations, prior doctor visits, self‐rated health status, informal caregiver, and comorbid coronary heart disease and diabetes mellitus.11, 12

Patient Exclusion Criteria

We excluded patients if the discharge destination was a nursing home, another acute care hospital, or an inpatient rehabilitation unit. Patients were excluded if life expectancy was less than 6 months as estimated by the hospitalist. Because follow‐up occurred via interview, patients without telephones or English‐ or Spanish‐language skills were excluded. Patients with cognitive impairment could participate with consent from a legally authorized representative and with a proxy whospent a minimum of 3 hours daily with the patient and was willing to answer postdischarge interviews.

Baseline Assessment

During the index hospital admission, trained data abstractors recorded baseline patient data: age, sex, race, diabetes mellitus, heart failure, chronic obstructive pulmonary disease, and coronary heart disease. Patients or proxies provided the number of hospital admissions and doctor visits during the year before the index hospital admission. We recorded the availability of an informal caregiver in response to the question Is there a friend, relative, or neighbor who would take care of you for a few days if necessary? Patients rated their health status on the following scale: poor, fair, good, very good, or excellent.

Discharge and Postdischarge Procedures

At the end of the index hospitalization, hospitalists and ward nurses used standardized forms for discharge diagnoses, prescriptions, instructions, and appointments. Discharge‐planning nurses or social workers consulted with hospitalists and ward nurses and then coordinated service providers including home health nurses, physical therapists, home health aides, homemaker service providers, durable medical equipment vendors, home oxygen vendors, home infusion pharmacists, social workers, rehabilitation service providers, legal aide providers, and others. After discharge, trained research personnel conducted 2 telephone interviews with the patient or the patient's proxy. The first interview occurred 1 week after discharge. Interviewers read verbatim items from the PREPARED10 and the Satisfaction with Information about Medicines Scale (SIMS).13 During the second telephone interview 30 10 days after discharge, interviewers recorded if patients had experienced at least 1 emergency department visit during the month after discharge.

The purpose of the PREPARED items was to have a bank of items and responses that could be used to generate the B‐PREPARED scale. The PREPARED questionnaire was originally developed to provide feedback to hospital ward staff about the quality of discharge‐planning activities that occurred during hospitalization.9 Discriminant factor analysis on the original 16 process questions revealed 4 factors that explained 57% of the total variance in patient/caregiver responses. The PREPARED domains included information exchange on community services and equipment, management of medication, the process of preparing to cope after discharge, and having control over one's discharge circumstances.9 The purpose of the SIMS was as a construct to compare with the B‐PREPARED scale. The derivation and validity of the SIMS have been described extensively elsewhere.13 In summary, the SIMS items were derived from recommendations of the Association of the British Pharmaceutical Industry. The intent of the SIMS was to determine if a patient's medication information needs were met and to allow comparison between patients or groups. Respondents selected 1 of 5 options for each of the 17 items. The sum of scores for each of the SIMS items yielded a total score that ranged from 0 to 17. Patients with high total SIMS scores had high satisfaction with the amount of medication information they received. Validation samples included inpatients and outpatients with a variety of diseases and characteristics. SIMS demonstrated adequate internal consistency, test‐retest reliability, and criterion‐related validity.13

Item Selection and Scoring of the B‐PREPARED Instrument

We selected an initial pool of items from the PREPARED instrument.10 The goal was a parsimonious, comprehensive, and valid instrument for use in clinical and research environments. When we retained or deleted items, our decision process was conservative, conceptual, and statistical. We performed item reduction in the following steps defined a priori. First, we agreed on items consistent with domains in the prepared for discharge construct as defined by expert consensus.9 Second, we excluded items that assessed qualities of the discharge process that were imperceptible to the patient on the day of discharge. Third, we excluded items that elicited open‐ended responses unsuitable for quantitative scale development and analysis. Fourth, we assessed reliability as defined by the Cronbach's alpha statistic. We excluded items that substantially decreased Cronbach's alpha.

Measures of Construct Validity

We used 2 measures of construct validity in our assessment of B‐PREPARED. One construct was patient worry. During the interview 1 week after discharge, research personnel asked, Now that you have been out of the hospital for a while, has anything been worrying you about managing at home? Response options for the dichotomous worry item were no or yes. We anticipated worried patients would have lower B‐PREPARED scale values. The other construct, SIMS, evaluated patient preparedness related to medication information exchange. The hypothesis was a positive correlation between SIMS and B‐PREPARED scale values.

Measure of Predictive Validity

We asked if B‐PREPARED predicted and discriminated groups of patients who did or did not visit emergency departments after hospital discharge. Emergency department visits were relevant adverse outcomes because of their association with postdischarge adverse events due to inpatient treatment.4 Emergency department visits reflected new or worsening symptoms after discharge. In our scale analysis, the hypothesis was patients with at least 1 emergency department visit would have lower B‐PREPARED scale values.

Analysis

Analyses were performed with SPSS PC (Version 14.0.2, SPSS Inc, Chicago, IL). We reported descriptive statistics as means, standard deviations, and range for interval variables and percentages for nominal variables. To determine the internal consistency of the scale, we calculated Cronbach's alpha. We assessed the distribution of the B‐PREPARED scale with visual and statistical tests for skewness. While using the SPSS FACTOR program, we performed principal components extractions and then rotated components using the oblique promax technique. Component scores were saved using the regression score procedure. Component loadings above 0.30 were considered important. Statistical inference tests were the Mann‐Whitney U for median differences between 2 groups and the Spearman correlation for associations. We reported medians with 25th and 75th percentiles. Differences between 2 correlations were tested using Fischer z transformations. The accepted level of significance was P < 0.05.

Description of Cohort

We approached 5124 patients during the index hospital admission. After applying exclusion criteria, we obtained consent and enrolled 491 patients. The reasons for exclusion were low Pra score for 34.9% of ineligible patients, discharge to nursing home for 12.8%, declined consent for 10.8%, nonparticipating hospitalist service for 9.1%, discharged during screen for 8.5%, previously enrolled in study for 5.6%, and declined screening for 2.3%. Each of the other exclusion criteria accounted for less than 4% of the ineligible patients. After subtracting 6% of eligible patients (31 of 491) who died, withdrew, or were lost during the first month, there were 460 patients available for analysis. Table 1 describes the patients' characteristics. Most of the patients, 75.2% (346 of 460), were less than 65 years old, and the mean age was 53.9 15.5 years. Many patients had chronic diseases including diabetes mellitus, coronary heart disease, heart failure, and chronic obstructive pulmonary disease. Most patients, 81.5% (375 of 460), rated their health as poor or fair, and 53.5% (246/460) had 1 or more hospital admissions during the year before their index admission. Cohort patients had a high probability of repeat admission: mean Pra 0.49 0.07 (range 0.400.70).

Item Reduction, Internal Consistency, and Score Distributions

Item reduction resulted in 12 items that fulfilled conceptual criteria. Table 2 shows the items and the distribution of responses. One of the 12 items, delays on the day you left the hospital (item 12, Table 2), was deleted because the item depressed the Cronbach's alpha. The B‐PREPARED with 11 items had acceptable internal consistency for the full cohort (Cronbach's alpha = 0.76).

For an individual patient, the sum of the scores for each item yielded a B‐PREPARED scale value. In the 460‐patient cohort, B‐PREPARED scale values had a mean of 17.3 4.3 and a negatively skewed distribution. A high scale value reflected high perception of discharge preparedness. Each of the 11 items correlated significantly with the B‐PREPARED scale value (P < 0.001, 2‐tailed).

There were substantial ceiling effects with individual items but not in the B‐PREPARED total score. Five of the 9 items with 3 response options had a ceiling effect above 70%. Three items had a dichotomous response option (items 7, 8, and 9). In 2 of these 3 items, more than 90% of respondents selected the response indicating higher preparedness. The total B‐PREPARED did not have noteworthy floor or ceiling effects. In this sample's total B‐PREPARED scores, 0.2% of respondents had the lowest score of 3, and 20% had the highest score of 22.

Principal Component Analysis

In the component analysis, we evaluated the correlation matrix of the 11 items in the B‐PREPARED scale. A Kaiser‐Meyer‐Olkin statistic of 0.76 indicated sufficient sampling adequacy to extract components from the matrix. Principal components extracted 54.2% of the variance associated with the 11‐item B‐PREPARED scale. After inspection of scree plots, we determined that 3 components were extracted before the eigenvalue fell below 1. The pattern matrix for the promax rotation was inspected, and the factor loading of each item appears in Table 3. The item content identified the first component as self‐care information for medications and activities. The second component was equipment and services. The third component was confidence. All B‐PREPARED items loaded primarily on 1 of the 3 components (Table 3).

Construct Validity

We assessed 2 constructs: worry and satisfaction with medication information (SIMS). In the cohort, 25% of patients (115 of 460) reported worry about managing at home. Worried patients had significantly lower B‐PREPARED scale values (median [25%, 75%] = 14 [10, 16]) than patients who did not worry (median [25%, 75%] = 17 [14, 20], P < .001). We calculated SIMS and then correlated SIMS with B‐PREPARED and components. In the cohort, the mean SIMS was 12.1 4.7 (range 0‐17). Patients with greater satisfaction on the SIMS also had higher B‐PREPARED scale values (rho = 0.45, P < 0.001). There was a significant positive correlation between SIMS and the B‐PREPARED component called self‐care information for medications and activities (rho = 0.46, P < 0.001). The other 2 B‐PREPARED components, equipment/services and confidence, were positively correlated with SIMS at much lower levels (rho = 0.18 and rho = 0.24, respectively, both Fischer z transformations P < .001). The B‐PREPARED scale demonstrated validity with the constructs of worry and satisfaction with medication information.

Predictive Validity

We assessed the capacity of the B‐PREPARED to predict and discriminate groups of patients who did or did not visit emergency departments. Within 1 month of hospital discharge, 16.5% of the cohort (76 of 460 patients) had at least 1 visit to an emergency department. B‐PREPARED scale values were lower for those patients who visited emergency departments (median [25%, 75%] = 14 [12, 18]) than those who did not (median [25%, 75%] = 16 [13, 19], P = 0.011). The B‐PREPARED scale analysis supported the hypothesized relationship with emergency department visits.

Correlations between B‐PREPARED and Baseline Characteristics

We evaluated the correlations between a patient's B‐PREPARED scale value and baseline characteristics, shown in Table 1. There was a weak positive correlation with self‐rated health status (rho 0.17, P < .001). Patients who perceived better health status had higher B‐PREPARED scale values than those with poorer status. The other baseline characteristics in Table 1 were not associated with B‐PREPARED scale values.

Patient Inclusion Criteria

Trained research coordinators identified all consecutive adult inpatients who were discharged to the patient's home by internal medicine hospitalists. Patient inclusion in our cluster randomized trial required a probability of repeat admission (Pra) score of at least 0.40.11, 12 Hence, the patients in the scale analysis cohort had the same high probability for repeat admission. The research coordinators calculated the Pra within 2 days before discharge from the index hospitalization. The Pra score came from a logistic model of age, sex, prior hospitalizations, prior doctor visits, self‐rated health status, informal caregiver, and comorbid coronary heart disease and diabetes mellitus.11, 12

Patient Exclusion Criteria

We excluded patients if the discharge destination was a nursing home, another acute care hospital, or an inpatient rehabilitation unit. Patients were excluded if life expectancy was less than 6 months as estimated by the hospitalist. Because follow‐up occurred via interview, patients without telephones or English‐ or Spanish‐language skills were excluded. Patients with cognitive impairment could participate with consent from a legally authorized representative and with a proxy whospent a minimum of 3 hours daily with the patient and was willing to answer postdischarge interviews.

Baseline Assessment

During the index hospital admission, trained data abstractors recorded baseline patient data: age, sex, race, diabetes mellitus, heart failure, chronic obstructive pulmonary disease, and coronary heart disease. Patients or proxies provided the number of hospital admissions and doctor visits during the year before the index hospital admission. We recorded the availability of an informal caregiver in response to the question Is there a friend, relative, or neighbor who would take care of you for a few days if necessary? Patients rated their health status on the following scale: poor, fair, good, very good, or excellent.

Discharge and Postdischarge Procedures

At the end of the index hospitalization, hospitalists and ward nurses used standardized forms for discharge diagnoses, prescriptions, instructions, and appointments. Discharge‐planning nurses or social workers consulted with hospitalists and ward nurses and then coordinated service providers including home health nurses, physical therapists, home health aides, homemaker service providers, durable medical equipment vendors, home oxygen vendors, home infusion pharmacists, social workers, rehabilitation service providers, legal aide providers, and others. After discharge, trained research personnel conducted 2 telephone interviews with the patient or the patient's proxy. The first interview occurred 1 week after discharge. Interviewers read verbatim items from the PREPARED10 and the Satisfaction with Information about Medicines Scale (SIMS).13 During the second telephone interview 30 10 days after discharge, interviewers recorded if patients had experienced at least 1 emergency department visit during the month after discharge.

The purpose of the PREPARED items was to have a bank of items and responses that could be used to generate the B‐PREPARED scale. The PREPARED questionnaire was originally developed to provide feedback to hospital ward staff about the quality of discharge‐planning activities that occurred during hospitalization.9 Discriminant factor analysis on the original 16 process questions revealed 4 factors that explained 57% of the total variance in patient/caregiver responses. The PREPARED domains included information exchange on community services and equipment, management of medication, the process of preparing to cope after discharge, and having control over one's discharge circumstances.9 The purpose of the SIMS was as a construct to compare with the B‐PREPARED scale. The derivation and validity of the SIMS have been described extensively elsewhere.13 In summary, the SIMS items were derived from recommendations of the Association of the British Pharmaceutical Industry. The intent of the SIMS was to determine if a patient's medication information needs were met and to allow comparison between patients or groups. Respondents selected 1 of 5 options for each of the 17 items. The sum of scores for each of the SIMS items yielded a total score that ranged from 0 to 17. Patients with high total SIMS scores had high satisfaction with the amount of medication information they received. Validation samples included inpatients and outpatients with a variety of diseases and characteristics. SIMS demonstrated adequate internal consistency, test‐retest reliability, and criterion‐related validity.13

Item Selection and Scoring of the B‐PREPARED Instrument

We selected an initial pool of items from the PREPARED instrument.10 The goal was a parsimonious, comprehensive, and valid instrument for use in clinical and research environments. When we retained or deleted items, our decision process was conservative, conceptual, and statistical. We performed item reduction in the following steps defined a priori. First, we agreed on items consistent with domains in the prepared for discharge construct as defined by expert consensus.9 Second, we excluded items that assessed qualities of the discharge process that were imperceptible to the patient on the day of discharge. Third, we excluded items that elicited open‐ended responses unsuitable for quantitative scale development and analysis. Fourth, we assessed reliability as defined by the Cronbach's alpha statistic. We excluded items that substantially decreased Cronbach's alpha.

Measures of Construct Validity

We used 2 measures of construct validity in our assessment of B‐PREPARED. One construct was patient worry. During the interview 1 week after discharge, research personnel asked, Now that you have been out of the hospital for a while, has anything been worrying you about managing at home? Response options for the dichotomous worry item were no or yes. We anticipated worried patients would have lower B‐PREPARED scale values. The other construct, SIMS, evaluated patient preparedness related to medication information exchange. The hypothesis was a positive correlation between SIMS and B‐PREPARED scale values.

Measure of Predictive Validity

We asked if B‐PREPARED predicted and discriminated groups of patients who did or did not visit emergency departments after hospital discharge. Emergency department visits were relevant adverse outcomes because of their association with postdischarge adverse events due to inpatient treatment.4 Emergency department visits reflected new or worsening symptoms after discharge. In our scale analysis, the hypothesis was patients with at least 1 emergency department visit would have lower B‐PREPARED scale values.

Analysis

Analyses were performed with SPSS PC (Version 14.0.2, SPSS Inc, Chicago, IL). We reported descriptive statistics as means, standard deviations, and range for interval variables and percentages for nominal variables. To determine the internal consistency of the scale, we calculated Cronbach's alpha. We assessed the distribution of the B‐PREPARED scale with visual and statistical tests for skewness. While using the SPSS FACTOR program, we performed principal components extractions and then rotated components using the oblique promax technique. Component scores were saved using the regression score procedure. Component loadings above 0.30 were considered important. Statistical inference tests were the Mann‐Whitney U for median differences between 2 groups and the Spearman correlation for associations. We reported medians with 25th and 75th percentiles. Differences between 2 correlations were tested using Fischer z transformations. The accepted level of significance was P < 0.05.

Description of Cohort

We approached 5124 patients during the index hospital admission. After applying exclusion criteria, we obtained consent and enrolled 491 patients. The reasons for exclusion were low Pra score for 34.9% of ineligible patients, discharge to nursing home for 12.8%, declined consent for 10.8%, nonparticipating hospitalist service for 9.1%, discharged during screen for 8.5%, previously enrolled in study for 5.6%, and declined screening for 2.3%. Each of the other exclusion criteria accounted for less than 4% of the ineligible patients. After subtracting 6% of eligible patients (31 of 491) who died, withdrew, or were lost during the first month, there were 460 patients available for analysis. Table 1 describes the patients' characteristics. Most of the patients, 75.2% (346 of 460), were less than 65 years old, and the mean age was 53.9 15.5 years. Many patients had chronic diseases including diabetes mellitus, coronary heart disease, heart failure, and chronic obstructive pulmonary disease. Most patients, 81.5% (375 of 460), rated their health as poor or fair, and 53.5% (246/460) had 1 or more hospital admissions during the year before their index admission. Cohort patients had a high probability of repeat admission: mean Pra 0.49 0.07 (range 0.400.70).

Item Reduction, Internal Consistency, and Score Distributions

Item reduction resulted in 12 items that fulfilled conceptual criteria. Table 2 shows the items and the distribution of responses. One of the 12 items, delays on the day you left the hospital (item 12, Table 2), was deleted because the item depressed the Cronbach's alpha. The B‐PREPARED with 11 items had acceptable internal consistency for the full cohort (Cronbach's alpha = 0.76).

For an individual patient, the sum of the scores for each item yielded a B‐PREPARED scale value. In the 460‐patient cohort, B‐PREPARED scale values had a mean of 17.3 4.3 and a negatively skewed distribution. A high scale value reflected high perception of discharge preparedness. Each of the 11 items correlated significantly with the B‐PREPARED scale value (P < 0.001, 2‐tailed).

There were substantial ceiling effects with individual items but not in the B‐PREPARED total score. Five of the 9 items with 3 response options had a ceiling effect above 70%. Three items had a dichotomous response option (items 7, 8, and 9). In 2 of these 3 items, more than 90% of respondents selected the response indicating higher preparedness. The total B‐PREPARED did not have noteworthy floor or ceiling effects. In this sample's total B‐PREPARED scores, 0.2% of respondents had the lowest score of 3, and 20% had the highest score of 22.

Principal Component Analysis

In the component analysis, we evaluated the correlation matrix of the 11 items in the B‐PREPARED scale. A Kaiser‐Meyer‐Olkin statistic of 0.76 indicated sufficient sampling adequacy to extract components from the matrix. Principal components extracted 54.2% of the variance associated with the 11‐item B‐PREPARED scale. After inspection of scree plots, we determined that 3 components were extracted before the eigenvalue fell below 1. The pattern matrix for the promax rotation was inspected, and the factor loading of each item appears in Table 3. The item content identified the first component as self‐care information for medications and activities. The second component was equipment and services. The third component was confidence. All B‐PREPARED items loaded primarily on 1 of the 3 components (Table 3).

Construct Validity

We assessed 2 constructs: worry and satisfaction with medication information (SIMS). In the cohort, 25% of patients (115 of 460) reported worry about managing at home. Worried patients had significantly lower B‐PREPARED scale values (median [25%, 75%] = 14 [10, 16]) than patients who did not worry (median [25%, 75%] = 17 [14, 20], P < .001). We calculated SIMS and then correlated SIMS with B‐PREPARED and components. In the cohort, the mean SIMS was 12.1 4.7 (range 0‐17). Patients with greater satisfaction on the SIMS also had higher B‐PREPARED scale values (rho = 0.45, P < 0.001). There was a significant positive correlation between SIMS and the B‐PREPARED component called self‐care information for medications and activities (rho = 0.46, P < 0.001). The other 2 B‐PREPARED components, equipment/services and confidence, were positively correlated with SIMS at much lower levels (rho = 0.18 and rho = 0.24, respectively, both Fischer z transformations P < .001). The B‐PREPARED scale demonstrated validity with the constructs of worry and satisfaction with medication information.

Predictive Validity

We assessed the capacity of the B‐PREPARED to predict and discriminate groups of patients who did or did not visit emergency departments. Within 1 month of hospital discharge, 16.5% of the cohort (76 of 460 patients) had at least 1 visit to an emergency department. B‐PREPARED scale values were lower for those patients who visited emergency departments (median [25%, 75%] = 14 [12, 18]) than those who did not (median [25%, 75%] = 16 [13, 19], P = 0.011). The B‐PREPARED scale analysis supported the hypothesized relationship with emergency department visits.

Correlations between B‐PREPARED and Baseline Characteristics

We evaluated the correlations between a patient's B‐PREPARED scale value and baseline characteristics, shown in Table 1. There was a weak positive correlation with self‐rated health status (rho 0.17, P < .001). Patients who perceived better health status had higher B‐PREPARED scale values than those with poorer status. The other baseline characteristics in Table 1 were not associated with B‐PREPARED scale values.