April 22, 2024
Health

Understanding inherent influencing factors to digital health adoption in general practices through a mixed-methods analysis


Adoption barriers and improvement strategies in general practices (literature review and expert interview results)

Our literature review and expert interviews aimed to identify and synthesize currently postulated adoption barriers and improvement strategies for digital health adoption more broadly and validate their relevance in general practice settings. We initially retrieved 1276 records in the literature search, of which we included 24 articles13,15,17,18,19,22,23,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44.

The literature review identified technological, social, and organizational adoption barriers. More than 90% of included studies reported organizational adoption barriers13,15,17,18,22,23,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44 (23/24), with more than half reporting high workload17,22,29,30,34,36,37,38,39,40,41,42,43,44 and a lack of time13,15,17,18,23,28,29,31,32,33,34,36,40,42 (each 14/24; 58%) as predominant barriers. Another 88% of studies identified social adoption barriers13,15,17,18,22,23,28,29,30,31,32,35,36,37,38,39,40,41,42,43,44 (21/24). Of these, physicians’ familiarity with digital health solutions15,17,18,22,23,28,30,31,32,36,38,39,40,41,42,43,44 (17/24; 71%) was the most cited social barrier, followed by overall awareness15,18,22,23,29,30,32,35,43,44 (10/24; 42%) and patient preferences15,18,23,29,30,31,35,38,40,42 (10/24; 42%).

Our ten expert interviews with GPs confirmed and validated the relevance of all three categories of barriers and five categories of improvement strategy. Overall, the relevance of the three categories of barriers was consistently rated as high. In line with the high estimated relevance, all GPs mentioned technological barriers, especially regarding system reliability (10/10; 100%), usefulness (9/10; 90%), and technical support (9/10; 90%). Additionally, most GPs mentioned the familiarity and ability of practice staff (each 8/10; 80%), patients’ preferences and ability (8/10; 80%), a lack of reimbursement (9/10; 90%), a high workload and lack of time (each 9/10; 90%), and the socio-political context (9/10; 90%) as substantial adoption barriers. On average, GPs reported around 14 different barriers.

Looking into potential strategies to support and improve digital health adoption, we identified strategies in five categories in our literature review: development-related, awareness-related, knowledge-related, implementation-related, and policy-related strategies. Around two-thirds of studies identified strategies concerning the development of digital health solutions as potentially helpful to improve adoption15,17,18,19,28,29,31,33,34,36,37,38,39,42,43,44 (16/24; 67%). Among these, the most frequently cited development-related strategies were improvements in the usefulness of digital health solutions17,18,19,28,29,31,33,36,37,38,42,43,44 (13/24; 54%), followed by improvements in their usability28,29,31,34,36,39,42,43 (8/24; 33%). All other categories were present in around half of the included studies, with the call for ongoing training15,17,18,19,34,37,38,40,43,44 (10/24; 42%) and improved reimbursement15,17,18,22,34,38,39,43 (8/24; 33%) as additionally vital improvement strategies.

Our expert interviews further highlighted that GPs considered development-related strategies particularly relevant: 80% of GPs would like to see improved usability of digital health solutions (8/10). In addition, GPs especially called for improvements in remuneration (8/10; 80%) and a simplification of political guidelines (9/10; 90%). Awareness-related strategies were rated as least relevant (6.2/10.0). Of these, GPs wished for further information on the functionalities and benefits of digital health solutions (each 7/10; 70%). Overall, GPs reported around 11 strategies. In our subsequent online survey, we only included items for barriers or strategies proposed by more than four articles or mentioned by more than one interviewee to ensure theoretical and expert consensus. Figure 1 shows the synthesized results.

Fig. 1: Overview of categories and individual barriers (strategies) based on the literature review and expert interview results.
figure 1

The figure shows categories and corresponding individual barriers (strategies) as well as their appearance in the literature review and expert interviews. nLR represents the number of studies identified in the literature review proposing the barrier (strategy); nEI shows the number of expert interviews in which the barrier (strategy) was mentioned. Light grey boxes with italic text show barriers (strategies) not included in the subsequent online survey. dhs digital health solutions.

Factors influencing adoption barriers and improvement strategies (online survey results)

To analyze factors that may influence adoption barriers and improvement strategies, our online survey focused on five areas of inherent characteristics: (i) demographics and practice-related characteristics, (ii) digital health usage, (iii) digital affinity, (iv) personality, and (v) digital maturity of the practice.

After data cleaning, quality, and privacy control, we included a broad sample of 216 German GPs with a diverse set of demographics (see Fig. 2).

Fig. 2: Characteristics of participating GPs (N = 216).
figure 2

The figure shows assessed individual and practice-related characteristics of participating GPs.

Around half of respondents used digital health solutions daily (93/216; 43.1%), while almost a third did not use them at all (20/216; 9.3%) or rather seldomly (47/216; 21.8%). Most respondents further expected to rather or very likely use digital health solutions in the future (161/216; 74.5%).

Further, GPs perceived the work-related digital affinity of their medical assistants to be moderate (55/216; 25.5%) or relatively high (91/216; 42.1%) and had a relatively moderate affinity for technology interaction45 themselves (M = 2.66, SD = 1.08).

Concerning personality46, respondents can be characterized as highly conscientious (M = 4.10, SD = 0.59) and open (M = 3.85, SD = 0.68), moderately extroverted (M = 3.64, SD = 0.80) and agreeable (M = 3.53, SD = 0.75), and mildly neurotic (M = 2.42, SD = 0.71). The digital maturity of their practices was moderate (M = 3.32, SD = 0.64).

Overall, respondents saw around 11 barriers (M = 11.12, SD = 6.01) and rated these as relatively moderate (M = 3.08, SD = 0.68). Among the three categories, organizational barriers were rated highest on average (M = 3.56, SD = 0.71), followed by technological (M = 2.93, SD = 0.76) and social (M = 2.76, SD = 0.79) barriers. For most individual barriers, scores were again moderate, with the highest rating for required workflow adjustments (M = 4.13, SD = 0.93), high costs and inadequate reimbursement (M = 4.02, SD = 1.02), and a high training and familiarization effort (M = 3.87, SD = 1.01) as the top three barriers (see Fig. 3).

Fig. 3: Sample size, mean, standard deviation, agreement rates, and between-group comparison for adoption barriers along the categories assessed.
figure 3

The figure shows items for adoption barriers per category, the respective sample size, descriptive statistics, and between-group comparison. The dot chart shows the mean value per item. Error bars represent +/− 2 standard errors. Cells with red framing show substantial differences between groups. %A Percentage of respondents agreeing to the statements and thus rating the respective barrier as relevant rating of (4) or (5); TB technological barriers; SB social barriers; OB organizational barriers; dhs digital health solutions.

On average, respondents perceived around 16 improvement strategies as important (M = 3.89, SD = 0.61). Policy-related (M = 4.00, SD = 0.81) and development-related strategies (M = 3.98, SD = 0.67) received the highest rating, followed by implementation-related (M = 3.90, SD = 0.78) and knowledge-related strategies (M = 3.85, SD = 0.81). Awareness-related strategies scored lowest but were also perceived as important (M = 3.70, SD = 0.74). Most individual strategies were similarly rated important (see Fig. 4): Respondents especially wished for improved interoperability (M = 4.38, SD = 0.81), continued technical support (M = 4.33, SD = 0.91), and improved usability (M = 4.20, SD = 0.88).

Fig. 4: Sample size, mean, standard deviation, agreement rates, and between-group comparisons for improvement strategies along the categories assessed.
figure 4

The figure shows items for improvement strategies per category, the respective sample size, descriptive statistics, and between-group comparisons. The dot chart shows the mean value per item. Error bars represent +/−2 standard errors. Cells with red framing show substantial differences between groups. %A Percentage of respondents agreeing to the statement and thus rating the respective strategy as important rating of (4) or (5); DS development-related strategies; AS awareness-related strategies; KS knowledge-related strategies; IS implementation-related strategies; PS policy-related strategies; dhs digital health solutions.

We conducted separate univariate ANOVAs and post hoc tests, to assess differences in the number and strength of adoption barriers and the number and importance of improvement strategies given the several inherent characteristics considered (see Fig. 5).

Fig. 5: Univariate ANOVAs and post hoc tests.
figure 5

Both parts of the figure show the results for Welch ANOVAs (left) and Hochberg GT2 or Games-Howell post hoc tests for significant Welch ANOVAs in the order of appearance (right). The upper part reports results for the strength of barriers, the lower part reports results for the importance of strategies. Blue brackets represent significant comparisons. As gender is a dichotomous variable, we conducted a two-tailed t-test. The results show the t-statistic (in the column ‘Welch’s F’), its’ df, and P-value. MA digital affinity medical assistants’ digital affinity; ATI affinity for technology interaction; N neuroticism; DM digital maturity.

The strength of barriers differed based on gender, current and future use of digital health solutions, GPs’ level of affinity for technology interaction, the level of extraversion and neuroticism, and the level of digital maturity. Post hoc tests revealed that participants who were female (M = 3.16, SD = 0.64, Cohen’s d = 0.25), never used digital health solutions (M = 3.42, SD = 0.64, p = 0.029, Cohen’s d = 0.73; Hochberg GT2 post hoc test), were very (M = 3.48, SD = 0.77, p = 0.034, Cohen’s d = 0.88; Hochberg GT2 post hoc test) or rather unlikely to use digital health solutions in the future (M = 3.61, SD = 0.59, p < 0.001, Cohen’s d = 1.12; Hochberg GT2 post hoc test), had a low level of affinity for technology interaction (M = 3.53, SD = 0.71), a low level of extraversion (M = 3.43, SD = 0.64, p = 0.011, Cohen’s d = 0.68; Hochberg GT2 post hoc test), a high (M = 3.63, SD = 0.60, p < 0.001, Cohen’s d = 1.04; Hochberg GT2 post hoc test) or moderate level of neuroticism (M = 3.21, SD = 0.61, p = .004, Cohen’s d = 0.44; Hochberg GT2 post hoc test), and a low (M = 3.43, SD = 0.56, p = 0.002, Cohen’s d = 0.90; Games-Howell post hoc test) or moderate level of digital maturity (M = 3.21, SD = 0.56, p < 0.001, Cohen’s d = 0.73; Games–Howell post hoc test) reported a higher strength of barriers compared to respondents who were male, used digital health solutions daily, were rather or very likely to use digital health solutions in the future, had a moderate or high level of affinity for technology interaction, a high level of extraversion, a low level of neuroticism, or a high level of digital maturity. Interestingly, male and female participants rated poor compatibility with work processes, a lack of reimbursement, high costs, and a high training effort as the most substantial adoption barriers (see Fig. 3).

We found a similar pattern for the number of barriers, except that there was no difference between GPs based on gender (t(214) = −1.397, p = 0.082; t-test), yet a significant difference based on the perceived digital affinity of medical assistants (Welch’s F(4, 21.51) = 3.433, p = .003; Welch ANOVA). GPs who perceived their medical assistants to be somewhat not digitally affine (M = 13.60, SD = 5.13) reported significantly more adoption barriers adoption compared to respondents perceiving their medical assistants to be fully digitally affine (M = 9.42, SD = 5.87, p = 0.053, Cohen’s d = 0.77; Hochberg GT2 post hoc test).

Looking at the importance of improvement strategies, we found significant differences between GPs based on gender, professional experience, current usage and expected future digital health usage, the level of neuroticism, and the level of digital maturity (see Fig. 5). Post hoc tests revealed that respondents who were female (M = 3.97, SD = 0.55, p = 0.017, Cohen’s d = 0.29; t-test), used digital health solutions daily (M = 3.95, SD = 0.62, p = 0.003, Cohen’s d = 0.87; Hochberg GT2 post hoc test), monthly (M = 4.00, SD = 0.54, p = 0.011, Cohen’s d = 0.97; Hochberg GT2 post hoc test), or seldomly (M = 3.95, SD = 0.59, p = .007, Cohen’s d = 0.88; Hochberg GT2 post hoc test), and were very (M = 3.96, SD = 0.59, p < 0.001, Cohen’s d = 1.25; Hochberg GT2 post hoc test) or rather likely to use digital health solutions in the future (M = 3.95, SD = 0.45, p = 0.002, Cohen’s d = 1.49; Hochberg GT2 post hoc test), reported a higher importance of strategies, compared to respondents who were male, never used digital health solutions, and were very unlikely to use these in the future. Interestingly, female participants viewed continuous technical support, improved interoperability, and improved reimbursement as the most vital improvement strategies. In contrast, for male participants, it was an enhanced interoperability, improved usefulness, and improved usability (see Fig. 4). Overall, we found similar results for the number of improvement strategies, except that there was an additional significant difference based on respondents’ level of conscientiousness (Welch’s F(2, 3.34) = 11.988, p = 0.030; Welch ANOVA).

In the next step, we conducted a linear hierarchical regression analysis to deepen our understanding of the association between adoption barriers, improvement strategies, and GPs’ inherent characteristics.

Looking at adoption barriers (see Table 1), demographics, practice-related characteristics, and digital health usage alone explained about 21.8% of the variance in the strength of barriers, reaching statistical significance of the model, F(21, 194) = 2.573, p < 0.001 (F-test). When including digital affinity variables in model 3, the proportion of explained criterion variance increase by 8.8% to an overall R2 of around 30.6% (F(23, 192) = 3.684, p < 0.001; F-test). Further including personality traits into our model led to an additional increase in R2 of 10.3%. Finally, also including digital maturity led to an increase in R2 of 3.6% to an overall R2 of 44.5% (F(29, 186) = 5.139, p < 0.001; F-test). Thus, our model significantly improved at each stage of the hierarchical process. The same was true for the number of adoption barriers, with a final R2 of 42.6% (F(29, 186) = 4.762, p = 0.005; F-test).

Table 1 Model parameters of the linear hierarchical regression model for the strength of barriers

In our final regression model, eight variables were significantly associated with the strength of barriers (see Supplementary Table 1 and Supplementary Table 2 for a detailed overview of coefficients). The strength of barriers was significantly associated with the practice location, the practice type, the expected future use of digital health solutions, GPs’ affinity for technology interaction, their extraversion, neuroticism, and openness, and the digital maturity of the practice. Accordingly, practicing in cities with less than 5,000 inhabitants compared to cities with 100,001 to 500,000 inhabitants (b = −0.315, SE B = 0.142, β = −0.152, p = 0.028; t-test) or cities with more than 500,000 inhabitants (b = −0.301, SE B = 0.133, β = −0.164, p = 0.025; t-test), sharing practices (b = 0.498, SE B = 0.189, β = 0.155, p = 0.009; t-test) compared to single practices, a lower expected likelihood of future usage (b = −0.151, SE B = 0.051, β = −0.281, p = 0.003; t-test), a lower affinity for technology interaction (b = −0.159, SE B = 0.042, β = −0.254, p < 0.001; t-test), lower extraversion (b = −0.109, SE B = 0.055, β = −0.129, p = 0.048; t-test), higher neuroticism (b = 0.156, SE B = 0.063, β = 0.164, p = 0.014; t-test) and openness (b = 0.134, SE B = 0.062, β = 0.135, p = 0.031; t-test), and lower digital maturity (b = −0.247, SE B = 0.071, β = −0.236, p < 0.001; t-test) were associated with a higher strength of barriers. We found similar results for the linear hierarchical regression model predicting the number of barriers, except that there was no substantial association with the practice type or extraversion.

Looking at the importance of improvement strategies (see Table 2), the model only including demographics and practice-related characteristics explained about 10.2% of the variance but did not reach statistical significance (F(16, 199) = 1.407, p = 0.141; F-test). Including digital health usage in our model yielded significant improvement in the proportion of explained criterion variance by 9.8%, leading to a total R2 of 20.0% (F(21, 194) = 2.305, p = 0.002; F-test). Further including digital affinity, personality, or digital maturity as predictors did not significantly improve the model, although the respective regression models were significant (see Table 2). Thus, the regression model only including demographics, practice-related characteristics, and digital health usage best fit our data.

Table 2 Model parameters of the linear hierarchical regression model for the importance of improvement strategies

In this model (model 2), three variables were significantly associated with the importance of improvement strategies. We found a significant association with respondents’ professional experience, their current usage of digital health solutions, and their expected future usage. Having 1 to 5 years of professional experience compared to 21 to 30 (b = −0.524, SE B = 0.210, β = −0.381, p = 0.013; t-test), using digital health solutions seldom (b = 0.458, SE B = 0.165, β = 0.308, p = 0.006; t-test) or monthly (b = 0.430, SE B = 0.208, β = 0.221, p = 0.040; t-test) compared to never, and a higher expected likelihood of future usage (b = 0.105, SE B = 0.052, β = 0.216, p = 0.043; t-test) were associated with a higher importance. Again, results were similar for the linear hierarchical regression model of the number of improvement strategies, except that there was an additional significant association with respondents’ age: Being aged between 46 and 55 (b = 3.682, SE B = 1.720, β = 0.302, p = 0.034; t-test) or older than 65 (b = 6.218, SE B = 2.849, β = .234, p = 0.030; t-test) was significantly associated with a higher number of strategies.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

SUBSCRIBE TO OUR NEWSLETTER

Get our latest downloads and information first.
Complete the form below to subscribe to our weekly newsletter.


100% secure your website.