Multi-level factors affecting timely electronic documentation of medication administration: a hierarchical linear modeling approach

Abstract

This study applies a systematic hierarchical linear modeling approach to identify factors impacting timely documentation of medication administration on electronic medication administration record (eMAR) systems. Delayed documentation of medications poses significant risks to patient safety. Multi-level quantitative data were collected from a large urban hospital system, spanning the non-physician clinician workforce across 27 patient-care units. Data suggests the overall perception of psychological safety on one’s unit was a significant predictor of individual clinicians’ timely eMAR documentation. The impact of each clinician’s personal psychological safety was nuanced by his/her patient-care unit and type of hospital. Other characteristics of the provider’s patient-care unit were also relevant. Thus, even though timely eMAR documentation is an individual-level activity, it is predicted by characteristics beyond complete control of the individual. We illustrate the value of applying systematic hierarchical linear modeling approach to better illuminate the problem of consistently achieving timely eMAR documentation across all providers.

APPENDIX

Model specifications for two levels of hierarchical linear model: effects of individual-level and unit-level factors, and their interaction effects, on the ‘proportion of timely eMAR documentations’

Level 1 model: Proportion of timely eMAR Documentations _ij =β _0j +β _1j *(Perceived Psychological Safety _ij )+β _2j *(Facilitating Conditions _ij )+β _3j *(Total Medications Administered _ij )+β _4j *(Gender _ij )+β _5j *(Technology Experience _ij )+β _6j *(Organizational Tenure _ij )+r _ij , where

β _0j :

Mean ‘proportion of timely eMAR documentations’ for providers within unit j

β _1j-6j :

Expected change in ‘proportion of timely eMAR documentations’ for person i in unit j associated with a unit increase in the corresponding Level 1 predictor variable characterizing person i within unit j

r _ij :

Error term representing a unique effect associated with person i within patient-care unit j, r _ij ~N (0, σ ²)

Level 2 models: (1) β _0j =γ ₀₀ +γ ₀₁ *(Unit Size _j )+γ ₀₂ *(Unit Type _j )+γ ₀₃ *(Unit Location _j )+γ ₀₄ *(Average Perceived Psychological Safety _j )+u _0j , where

γ ₀₀ :

Mean ‘proportion of timely eMAR documentations’ for a person across all patient-care units

γ ₀₁ :

Mean ‘proportion of timely eMAR documentations’ for a person located within a patient-care unit whose ‘Unit Size’ is the same as the average ‘unit size’ across all units

γ ₀₂ :

Difference in mean ‘proportion of timely eMAR documentations’ for a person working in a general care unit vs an ICU

γ ₀₃ :

Difference in mean ‘proportion of timely eMAR documentations’ for a person working in a patient-care unit that is located within a teaching vs a non-teaching hospital setting

γ ₀₄ :

Mean ‘proportion of timely eMAR documentations’ for a person located within a patient-care unit whose average ‘perceived psychological safety’ is the same as the average ‘perceived psychological safety’ across all units

u _0j :

The unique effect of patient-care unit j on mean ‘proportion of timely eMAR documentations’ holding all Level 2 indicators constant (i.e., conditioning on all Level 2 predictors)

(2) β _1j =γ ₁₀ +γ ₁₁ *(Unit Type _j )+γ ₁₂ *(Unit Size _j )+γ ₁₃ *(Unit Location _j )+u _1j , where

γ ₁₀ :

Mean ‘perceived psychological safety’ – ‘proportion of timely eMAR documentations’ slope for a person across all patient-care units

γ ₁₁ :

Difference in mean ‘perceived psychological safety’ – ‘proportion of timely eMAR documentations’ slope for a person working in a general care unit vs an ICU

γ ₁₂ :

Mean ‘perceived psychological safety’ – ‘proportion of timely eMAR documentations’ slope for a person located within a patient-care unit whose ‘unit size’ is the same as the average ‘unit size’ across all units

γ ₁₃ :

Difference in mean ‘perceived psychological safety’ – ‘proportion of timely eMAR documentations’ slope for a person working in a patient-care unit that is located within a teaching vs a non-teaching hospital setting

u _1j :

The unique effect of patient-care unit j on the ‘perceived psychological safety’ – ‘proportion of timely eMAR documentations’ slope holding all Level 2 indicators constant (i.e., conditioning on all Level 2 predictors)

(3–7) β _yj =γ _y0 +u _yj , where y=2-6,

γ _y0 :

Average slope of Level 1 indicator variable _y and ‘proportion of timely eMAR documentations’ across all patient-care units

u _yj :

The unique effect of patient-care unit j on the ‘Level 1 indicator variable _y – “proportion of timely eMAR documentations” ’ slope holding all Level 2 indicators constant (i.e., conditioning on all Level 2 predictors)

Interpretation of effect sizes:

The mean ‘perceived psychological safety’ that existed within the entire unit had the largest effect on individuals’ proportion of timely eMAR documentations. A unit increase in the overall ‘perceived psychological safety’ of all members, on average, within the unit was associated with a 3.11 units increase in the proportion of timely eMAR documentations by individual members within that patient-care unit. The Unit-level Effect in this model was also large (Effect Size: 0.99). Among the moderating relationships, the type of unit in which the individual was located had quite a large effect (0.33) on the relationship between the individual’s ‘perceived psychological safety’ and his/her proportion of timely eMAR documentations.

Table A1

Table A1 Convergent and discriminant validity measures

Table A2

Table A2 Effect sizes of hypothesized relationships