A measure of dignified treatment for healthcare workers: design and psychometric properties

Abstract

Despite the importance of both user and healthcare worker considerations to the achievement of a comprehensive understanding of dignified care in organizations, most studies in this regard have mainly considered patient aspects. This study aims to address this gap by designing and psychometrically validating a scale of organizational dignified treatment (ODT) toward healthcare workers. The sample consists of 442 healthcare workers from Colombia. We use a psychometric analytics framework comprising six processes that incorporate exploratory factor analysis, structural equation modeling, and machine learning methods. The final ODT scale contains six items and satisfies the requirements for reliability (Cronbach’s alpha, omega total, and composite reliability), goodness-of-fit, convergent and discriminant validity, equity (examining gender, age, education, and socioeconomic stratum using confirmatory factor analysis with control factors, differential item functioning, and satisficing analysis), criterion validity (considering healthcare worker well-being as the criterion), and predictive validity. Moreover, we identify and discuss six essential facets of ODT toward healthcare workers, namely, amiability, explained procedures, enjoyment in the workplace, security in the workplace, understandable information, and voice of healthcare workers.

Appendices

Appendix Teaching–learning resources for reproducing some results of the manuscript in R

Appendix 1 “Explore” process (Sect. 4.2)

library(psych)

# r.cor: Pearson’s correlation matrix (see Table 2).

n = 442.

cortest.bartlett(r.cor, n = n).

KMO(r.cor).

efa1 < -fa(r.cor,n.obs = n, nfactors = 1, fm = "pa", rotate = "varimax", scores = "regression").

efa1 < -round(cbind(efa1$Structure),3).

efa1[efa1 < 0.45] <-NA.

efa1.

psych::omega(r.cor, n.obs = 442, plot = F)$omega.tot.

psych::alpha(r.cor, n.obs = 442)$total[1].

#Note: (r.cor) only has three decimals. The resulting values may vary slightly in decimals concerning the reported in the manuscript.

Appendix 2 The refined CFA model (see Table 5 and Fig. 2)

Input: r.cov: Covariance matrix of the refined items.

	rt1	rt2	rt3	rt4	rt5	rt6
rt1	0.375	0.289	0.308	0.284	0.268	0.212
rt2	0.289	0.676	0.408	0.437	0.43	0.338
rt3	0.308	0.408	0.641	0.495	0.447	0.353
rt4	0.284	0.437	0.495	0.636	0.465	0.372
rt5	0.268	0.43	0.447	0.465	0.632	0.368
rt6	0.212	0.338	0.353	0.372	0.368	0.696

library(lavaan)

refined.model <-'ODT =  ~ rt1 + rt2 + rt3 + rt4 + rt5 + rt6'.

Mod.6.items <-cfa(refined.model, std.lv = T, sample.cov = r.cov, sample.nobs = 442, likelihood = "wishart", estimator = "ML").

summary(Mod.6.items, standardized = T, fit.measures = T).

#Note: r.cov only has three decimals. The resulting values may vary slightly in decimals concerning the reported in the manuscript.

Appendix 3 Illustrating DIF analysis for “rt1” and “rt2” (Table 7)

library(lordif)

dat.dif = read.csv("dat_dif.csv", header = T, sep = ",").

summary(dat.dif).

group < -dat.dif$gender.

tot = dat.dif$tot.

w < -rundif(item = 1:2, resp = dat.dif[,c(1:2)], theta = tot, gr = group, criterion = "CHISQR", alpha = 0.01, pseudo.R2 = "McFadden").

w$stats[,c(1:5,7:9)].

#Note: Table 2 download “dat_dif.csv” from https://doi.org/10.13140/RG.2.2.20883.55843

# “tot” is the sum of scores for each participant by considering all six items [In R: tot = rowSums(Data)].

Appendix 4 Illustrating satisficing analysis for “Lmax.bin” (Table 8)

dat.dif = read.csv("dat_dif.csv", header = T, sep = ",").

dat.satisf < -read.csv("dat_satisf.csv", header = T, sep = ",").

dat.satisf < -cbind(dat.satisf[, c(1:2)], dat.dif[, c(4:7)]).

summary(dat.satisf).

mod1 < -glm(Lmax.bin ~ gender + age + stratum + postg, family = binomial, data = dat.satisf).

print(summary(mod1), digits = 2).

#Note: download “dat_satisf.csv” from https://doi.org/10.13140/RG.2.2.25916.72329