In the Multi-State Nursing Care and Patient Safety Study, nurse survey data were collected in from nearly

Size: px

Start display at page:

Download "In the Multi-State Nursing Care and Patient Safety Study, nurse survey data were collected in from nearly"

Paul Baker
5 years ago
Views:

1 Stimpfel AW, Sloane DM, Aiken LH. The longer the shifts for hospital nurses, the higher the levels of burnout and patient dissatisfaction. Health Aff (Millwood). 2012;31(11). Technical Appendix METHODS--Design and Data Nurse Survey In the Multi-State Nursing Care and Patient Safety Study, nurse survey data were collected in from nearly 100,000 nurses in California, Pennsylvania, New Jersey, and Florida using state licensure lists as sampling frames. The survey was fairly wide ranging and included questions about among many other things the nurses shift length, work environment, staffing, burnout, job satisfaction, intentions to leave and characteristics of scheduling. Mailed surveys were sent directly to the homes of nurses, who represented random samples of 40% of all nurses licensed in California and Pennsylvania, 50% of all nurses licensed in New Jersey, and 25% of all licensed nurses in Florida. The different sampling fractions in the different states were driven by funding constraints. We used a modified Dillman approach with postcard reminders and a second mailing of questionnaires to nonrespondents. 1 The nurses surveyed included nurses working in all settings, in and out of direct patient care, and nurses who had chosen to maintain their licenses but not work as nurses, and some nurses who were unemployed.

2 The response rate was 39%. Response rates in large surveys have been decreasing for many years and a higher response rate may not always provide unbiased samples. 2,3 To determine the degree to which our sample was biased on the variables of proximal interest, we drew a random sample of nurses from the non-responders to the original survey from Pennsylvania (n=650) and California (n=650). Additional efforts were employed to avoid non-response from the smaller sample (e.g. phone calls, priority mail, and cash incentives) and as a result we achieved a 91% response rate in the sample of non-respondents. There were a few demographic differences between respondents and the original non-respondents, however, there was no evidence that the two samples differed with respect to nurses reports of burnout, job dissatisfaction, or other relevant variables in our study. 4 Due to the large number of respondents and the lack of evidence of response bias, we are confident in the reliability and validity of the findings from the survey. Additional Detail for Measures Burnout. Burnout was measured by the Emotional Exhaustion subscale of the Maslach Burnout Inventory that sums nine items, each measured by a seven-point ordinal scale, to assess nurses feelings of job-related emotional exhaustion. 5 Nurses were classified as being burned out if their score on the Emotional

3 Exhaustion subscale was higher than the published average greater than or equal to 27 for health care workers. 6 Job Satisfaction. Job satisfaction was measured on a fourpoint ordinal scale with responses ranging from very satisfied to very dissatisfied. For analysis purposes, job satisfaction was dichotomized to compare nurses who reported being either a little or very dissatisfied with nurses who reported being very or moderately satisfied. Intent to leave. Intent to leave was measured using a yes/no response to the question, Do you plan to be with your current employer one year from now? Unit specialty. A dichotomous variable was created from the nurse survey to categorize nurses working in intensive care units versus general care units. This was done in order to adjust for the differential influence that hospitals with more intensive care settings, resulting in better staffing ratios, would have on these models. The Practice Environment Scale of the Nursing Work Index. The instrument consists of five subscales related to (1) nurse participation in hospital affairs; (2) nursing foundations for quality of care; (3) nurse manager ability, leadership, and support of nurses; (4) staffing and resource adequacy; and (5) collegial nurse physician relations. As in previous research 7, the staffing and resource adequacy subscale (Staffing and

4 Resource Adequacy) was omitted because of its high correlation with our nurse staffing measure. For each hospital, we calculated subscale measures by averaging the values of all items in the subscale for all the nurses in the hospital. We then created a categorical summary measure for each hospital that has demonstrated good predictive validity where hospitals above the median on 3 or 4 subscales were classified as having good work environments; hospitals above the median on 1 or 2 subscales were classified as having mixed work environments; and hospitals above the median on zero subscales were classified as having poor work environments. Hospital Controls. Additional variables were derived from the nurse survey and American Hospital Association Annual Survey data to serve as control variables in the predictive models. Hospital characteristics were obtained from the American Hospital Association Annual Survey and included hospital size, teaching status, ownership, and core based statistical area, a measure of population density. These variables were included in the regression models to adjust for hospital factors that were related to the voluntary participation in the Hospital Consumer Assessment of Healthcare Providers and Systems survey during the study period prior to participation. Additional Detail for Analysis

5 For the descriptive analyses, frequencies, including numbers and percentages, were calculated for the major outcomes of interest. We used bivariate generalized estimating equation (GEE) models to assess the relationship between shift length and the three nurse outcomes. Next, multivariate generalized estimating equation (GEE) methods were used to reassess these relationships after controlling for potential confounds. The models we used account for the clustering of the nurses within hospitals 8 and produce robust standard error estimates that adjust for the correlations among clustered observations. Parameter estimates were transformed to odds ratios for ease of interpretation. The second part of the analysis entailed hospital-level regression modeling of patient satisfaction, and included all acute care hospitals in California, Florida, New Jersey and Pennsylvania for which we had nurse survey data and that reported Hospital Consumer Assessment of Healthcare Providers and Systems survey data to the Centers for Medicare and Medicaid Services. The final hospital sample for the regression models was 396. Specifically, ordinary least squares regression models were used to estimate the effect of nurse shift length on patient satisfaction. We first estimated bivariate models for each of the ten Hospital Consumer Assessment of Healthcare Providers and Systems survey outcomes and then estimated

6 multivariate models for the same outcomes that accounted for nurse age, hospital state, core-based statistical area, bed size, ownership, teaching status, practice environment, and nurse-to-patient ratio. Statistical significance was set at the.05 level using two-tailed tests for all analyses. References 1. Dillman DA. Mail and Internet Surveys the Tailored Design Method. New York: John Wiley & Sons; Asch DA, Jedrziewski MK, Christakis NA. Response rates to mail surveys published in medical journals. J Clin Epidemiol. 1997;50(10): PMID: Johnson TP, Wislar JS. Response rates and nonresponse errors in surveys. JAMA : the journal of the American Medical Association. 2012;307(17): Smith HL. A double sample to minimize bias due to nonresponse in a mail survey In: Ruiz-Gazen A, Guilbert P, Haziza D, Tille Y, eds. Paris: Dunod; 2008: Maslach C, Jackson SE. The measurement of experienced burnout. Journal of Organizational Behaviour. 1981;2: Maslach C, Jackson SE. Maslach Burnout Inventory Manual. 2nd ed. Palo Alto, CA: Consulting Psychologists Press; 1986.

7 7. McHugh MD, Kutney-Lee A, Cimiotti JP, Sloane DM, Aiken LH. Nurses' widespread job dissatisfaction, burnout, and frustration with health benefits signal problems for patient care. Health Aff (Millwood). 2011;30(2): PMID: Hanley JA, Negassa A, Edwardes MD, Forrester JE. Statistical analysis of correlated data using generalized estimating equations: An orientation. Am J Epidemiol. 2003;157(4): PMID:

8 Supplemental Exhibit 3 Full specifications of regression models to correspond with results discussing the relationship between nurses' shift length and nurse outcomes: burnout, job dissatisfaction and intention to leave Unadjusted Adjusted Dependent variable Covariates Estimate SE 95% CI p value Estimate 95% CI SE p value Burnout hour shift < < hour shift > 13 hour shift < < years old years old years old years old <.0001 Male ICU <.0001 Nurse staffing Mixed practice environment <.0001 Good practice environment < beds Beds Minor teaching Major teaching Technology Note. Number of observations=19,462; number of events (burnout)=6,763

9 Job dissatisfaction hour shift < < hour shift > 13 hour shift < < years old years old < years old < years old Male ICU Nurse staffing <.0001 Mixed practice environment <.0001 Good practice environment < beds Beds Minor teaching Major teaching Technology Note. Number of observations=19,274; number of events (dissatisfied)=5,082 Intention to leave hour shift hour shift < <.0001 > 13 hour shift < < years old < years old < years old < years old

10 Male <.0001 ICU Nurse staffing Mixed practice environment <.0001 Good practice environment < beds Beds Minor teaching Major teaching Technology Note. Number of observations=19,503; number of events (intent to leave)=2,846 The reference groups include: 8-9 hour shift category, 30 year old, female, non-icu, poor practice environment, <100 beds, non-teaching hospitals, and low technology.

11 Supplemental Exhibit 4 Full specifications of Regression models to correspond with results discussing the relationship between nurses' shift length and patient satisfaction using the HCAHPS, Hospital proportion of shifts above 13 hours Unadjusted Adjusted Dependent variable Covariates Estimate SE p value Estimate SE p value Do not recommend hospital Greater than 13 hours < Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R 2 =.0514, F(1, 395)= 21.35, p<.0001; Adjusted model R 2 =.2885, F(9, 358)=16.13 p<.0001 RN communication Greater than 13 hours < Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status

12 Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0482, F(1, 395)= 19.94, p<.0001; Adjusted model R2 =.2348, F(9, 358)=12.21 p<.0001 Quiet at night sometimes or never Greater than 13 hours Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state <.0001 Nurse staffing Practice environment Note: Unadjusted model R2 =.0077, F(1, 395)= 3.07, p=.0805; Adjusted model R2 =.17, F(9, 358)=8.15 p<.0001 Pain controlled sometimes or never Greater than 13 hours < Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0554, F(1, 395)= 23.12, p<.0001; Adjusted model R2 =.1987, F(9, 358)=9.86 p<.0001

13 Physician communication Greater than 13 hours Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0344, F(1, 395)= 14.05, p=.0002; Adjusted model R2 =.1339, F(9, 358)=6.15 p<.0001 Low rating of hospital Greater than 13 hours < Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R2 =.0575, F(1, 395)= 25.08, p<.0001; Adjusted model R2 =.2661, F(9, 358)=14.42 p<.0001 Help when needed sometimes or never Greater than 13 hours < Bed size Age

14 Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0505, F(1, 395)= 20.97, p<.0001; Adjusted model R2 =.2245, F(9, 358)=11.52 p<.0001 Explain medications sometimes or never Greater than 13 hours < Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0585, F(1, 395)= 24.49, p<.0001; Adjusted model R2 =.2864, F(9, 358)=15.97 p<.0001 Clean room sometimes or never Greater than 13 hours < Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing

15 Practice environment Note: Unadjusted model R2 =.0384, F(1, 395)= 15.73, p<.0001; Adjusted model R2 =.1968, F(9, 358)=9.75 p<.0001 Discharge information Greater than 13 hours Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0122, F(1, 395)= 4.88, p=.0278; Adjusted model R2 =.2097, F(9, 358)=10.44 p<.0001 Hospital proportion of shifts hours Unadjusted Adjusted Dependent variable Covariates Estimate SE p value Estimate SE p value Do not recommend hospital hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status

16 Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R 2 =.0256, F(1, 395)= 10.34, p=.0014; Adjusted model R 2 =.2882, F(9, 358)=16.1 p<.0001 RN communication hour shifts Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0398, F(1, 395)= 16.33, p<.0001; Adjusted model R2 =.2358, F(9, 358)=12.46 p<.0001 Quiet at night sometimes or never hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state <.0001 Nurse staffing Practice environment Note: Unadjusted model R2 =.0003, F(1, 395)= 1.12, p=.291; Adjusted model R2 =.172, F(9, 358)=8.27 p<.0001

17 Pain controlled sometimes or never hour shifts Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0188, F(1, 395)= 7.55, p<.0063; Adjusted model R2 =.1863, F(9, 358)=9.11 p<.0001 Physician communication hour shifts < Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0552, F(1, 395)= 23.03, p<.0001; Adjusted model R2 =.164, F(9, 358)=7.8 p<.0001 Low rating of hospital hour shifts Bed size Age

18 Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R2 =.011, F(1, 395)= 4.4, p=.0366; Adjusted model R2 =.2568, F(9, 358)=13.75 p<.0001 Help when needed sometimes or never hour shifts Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0343, F(1, 395)= 14, p=0002; Adjusted model R2 =.2074, F(9, 358)=10.41 p<.0001 Explain medications sometimes or never hour shifts Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing

19 Practice environment Note: Unadjusted model R2 =.0325, F(1, 395)= 13.25, p=.0003; Adjusted model R2 =.2861, F(9, 358)=15.94 p<.0001 Clean room sometimes or never hour shifts < <.0001 Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0715, F(1, 395)= 30.36, p<.0001; Adjusted model R2 =.2296, F(9, 358)=11.85 p<.0001 Discharge information hour shifts < <.0001 Bed size Age Core based statistical area <.0001 Ownership Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R2 =.0559, F(1, 395)= 23.34, p<.0001; Adjusted model R2 =.2405, F(9, 358)=12.6 p<.0001

20 Hospital proportion of shifts hours Unadjusted Adjusted Dependent variable Covariates Estimate SE p value Estimate SE p value Do not recommend hospital hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R 2 =.018, F(1, 395)= 7.23, p=.0075; Adjusted model R 2 =.2921, F(9, 358)=16.41 p<.0001 RN communication hour shifts Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001

21 Note: Unadjusted model R2 =.0329, F(1, 395)= 13.39, p=.0003; Adjusted model R2 =.2398, F(9, 358)=12.55 p<.0001 Quiet at night sometimes or never hour shifts Bed size Age Core based statistical area Ownership Teaching status Hospital state <.0001 Nurse staffing Practice environment Note: Unadjusted model R2 =.0234, F(1, 395)= 9.46, p=.0022; Adjusted model R2 =.1727, F(9, 358)=8.3 p<.0001 Pain controlled sometimes or never hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0164, F(1, 395)= 6.59, p=.0106; Adjusted model R2 =.1915, F(9, 358)=9.42 p<.0001 Physician communication hour shifts Bed size

22 Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0295, F(1, 395)= 11.98, p=.0006; Adjusted model R2 =.1534, F(9, 358)=7.21 p<.0001 Low rating of hospital hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R2 =.0083, F(1, 395)= 3.28, p=.0708; Adjusted model R2 =.2625, F(9, 358)=14.16 p<.0001 Help when needed sometimes or never hour shifts Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state

23 Nurse staffing Practice environment Note: Unadjusted model R2 =.0186, F(1, 395)= 7.47, p=0066; Adjusted model R2 =.2013, F(9, 358)=10.02 p<.0001 Explain medications sometimes or never hour shifts Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R2 =.0118, F(1, 395)= 4.7, p=.0307; Adjusted model R2 =.2828, F(9, 358)=15.69 p<.0001 Clean room sometimes or never hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0409, F(1, 395)= 16.8, p<.0001; Adjusted model R2 =.2113, F(9, 358)=10.66 p<.0001

24 Discharge information hour shifts < Bed size Age Core based statistical area <.0001 Ownership Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R2 =.045, F(1, 395)= 18.57, p<.0001; Adjusted model R2 =.2353, F(9, 358)=12.24 p<.0001 Hospital proportion of shifts 8-9 hours Unadjusted Adjusted Dependent variable Covariates Estimate SE p value Estimate SE p value Do not recommend hospital 8-9 hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R 2 =.0343, F(1, 395)= 14.01, p=.0002; Adjusted model R 2 =.2883, F(9, 358)=16.11 p<.0001

25 RN communication 8-9 hour shifts < Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0447, F(1, 395)= 18.44, p<.0001; Adjusted model R2 =.2387, F(9, 358)=12.48 p<.0001 Quiet at night sometimes or never 8-9 hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state <.0001 Nurse staffing Practice environment Note: Unadjusted model R2 =.0109, F(1, 395)= 4.35, p=.0377; Adjusted model R2 =.1731, F(9, 358)=8.33 p<.0001 Pain controlled sometimes or never 8-9 hour shifts Bed size Age

26 Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0266, F(1, 395)= 10.79, p=.0011; Adjusted model R2 =.1875, F(9, 358)=9.18 p<.0001 Physician communication 8-9 hour shifts < Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0605, F(1, 395)= 25.38, p<.0001; Adjusted model R2 =.1612, F(9, 358)=7.46 p<.0001 Low rating of hospital 8-9 hour shifts Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing

27 Practice environment <.0001 Note: Unadjusted model R2 =.0198, F(1, 395)= 7.97, p=.005; Adjusted model R2 =.2572, F(9, 358)=13.77 p<.0001 Help when needed sometimes or never 8-9 hour shifts < Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0416, F(1, 395)= 17.1, p<.0001; Adjusted model R2 =.2118, F(9, 358)=10.69 p<.0001 Explain medications sometimes or never 8-9 hour shifts < Bed size Age Core based statistical area <.0001 Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0449, F(1, 395)= 18.5, p<.0001; Adjusted model R2 =.2875, F(9, 358)=16.05 p<.0001

28 Clean room sometimes or never 8-9 hour shifts < <.0001 Bed size Age Core based statistical area Ownership <.0001 Teaching status Hospital state Nurse staffing Practice environment Note: Unadjusted model R2 =.0749, F(1, 395)= 31.88, p<.0001; Adjusted model R2 =.2285, F(9, 358)=11.78 p<.0001 Discharge information 8-9 hour shifts < Bed size Age Core based statistical area <.0001 Ownership Teaching status Hospital state Nurse staffing Practice environment <.0001 Note: Unadjusted model R2 =.0472, F(1, 395)= 19.53, p<.0001; Adjusted model R2 =.2333, F(9, 358)=12.1 p<.0001

THE CRITICAL CARE WORK ENVIRONMENT HEALTH CARE ASSOCIATED INFECTIONS AND NURSE-REPORTED. Patient Safety Issues. 1.0 Hour

Patient Safety Issues THE CRITICAL CARE WORK ENVIRONMENT AND NURSE-REPORTED HEALTH CARE ASSOCIATED INFECTIONS By Deena Kelly, RN, PhD, Ann Kutney-Lee, RN, PhD, Eileen T. Lake, RN, PhD, and Linda H. Aiken,