A randomized controlled pilot study investigating adherence to blood pressure diaries with personal pictures in stroke follow-up care

Stroke is the second leading cause of death worldwide and a major cause of permanent disability [1]. The increasing incidence of stroke and resulting death or disability is partly attributable to modifiable lifestyle factors [2]. Therefore, secondary stroke prevention is an integral part of stroke management. Arterial hypertension is a key modifiable risk factor in stroke prevention. Blood pressure (BP) lowering yields a 25–30% risk reduction for stroke recurrence [3, 4]. High BP is the most significant contributor to disability-adjusted life years in stroke [2].

Accurate BP measurement is a prerequisite for an effective antihypertensive therapy and BP monitoring is recommended twice daily [5]. Due to the impracticality of longitudinal clinic-based BP measurements, home monitoring of BP recorded in blood pressure diaries (BPDs) has gained popularity [6]. The affordability, convenience and patient empowerment along with its potentially higher prognostic value have driven its adoption. [7]. The BP home monitoring is feasible in people with poor BP control and the impact on secondary prevention is well-documented [8, 9]. Home-monitored BP reduces arterial BP, regardless of changes in antihypertensive therapy and had more favorable cardiovascular outcomes than BP measured in the hospital [8, 10].

However, fostering a routine for regular home BP measurements remains challenging. Studies on the adherence of patients to BP home measurements are scarce, which may obscure the potential to optimize secondary stroke prevention. Moreover, the lack of data precludes sample size calculations for studies investigating ways of boosting adherence to BPDs (i.e., improving the prognostic accuracy of recorded BP values). Addressing these shortcomings, we designed a randomized controlled pilot study to investigate whether the proper maintenance of BPDs is enhanced by the inclusion of pictures individually selected by patients for their personally meaningful and positive emotional content.

We report categorical variables as absolute frequencies and percentages and continuous variables as median ± interquartile range (IQR) or as mean ± standard deviation (SD) as appropriate. As a pilot study, our study did not include a priori hypotheses or a sample size calculation. Statistical tests were performed to check for underlying distributions and patient characteristic balance. Categorial variables were tested with the χ²-test or (for cases of expected absolute frequencies < 5) Fisher’s exact test. We tested continuous variables for normal distribution by the Shapiro-Wilk test. Between-group differences in continuous variables were tested with an unpaired two-sample two-tailed t‑test or Mann-Whitney U test as appropriate. Variances were tested with an F‑test. A p-value of < 0.05 was considered significant. There were no multiple comparisons. We performed randomization with an online tool (www.​randomizer.​at) [16]. In cases of missing values for PROs, we report the fraction of missing values and, where applicable, applied the personal mean score imputation method for completion [17]. Statistical analysis was performed using SPSS 21.0 (SPSS Inc, Chicago, IL, USA).

The BP home measurements provide high prognostic accuracy for assessing stroke risk, making BPDs valuable tools in secondary stroke prevention [18, 19]; however, their effectiveness in guiding treatment depends on patient adherence, highlighting the importance of studying factors that influence BPD adherence in people at risk for stroke. Earlier studies focused on other aspects of BPDs (e.g. self-titration of antihypertensives) and used coarse measures of BPD adherence [20‐25] or were not conducted in people with neurovascular diagnoses [12, 26]. Moreover, these studies did not assess adherence variability, preventing sample-size calculations for interventional studies. Our pilot study addressed these gaps, by investigating adherence to BPDs with personal, positive images provided by patients as well as patient preferences in their use of BPDs.

At roughly 60%, overall adherence to home BP measurements was suboptimal, emphasizing the need for interventions to motivate people with stroke or TIA to self-monitoring. Considering the Hawthorne effect that an observation itself influences behavior, the mean completion rate may even constitute an overestimation [27]. The integration of personalized visuals into BPDs led to a trend of higher adherence in our intervention group and may therefore hold potential to address low BPD adherence; however, this trend was not statistically significant. Therefore, our study does not provide evidence that personalized visuals improve BPD adherence, although the small sample size may have limited our ability to detect an effect. Earlier studies measured adherence to BPDs by counting time intervals (weeks or days) containing at least one entry but this approach overlooks BP variability within the defined time interval [21‐23]. Furthermore, as we consider each additionally available BP value to be clinically meaningful, a mere count of days with at least one entry may prevent identification of arterial hypertension in some patients. By this definition, one study reported a BPD adherence of nearly 100% [22]. Applying the same definition of adherence on our data would lead to greater estimates of adherence (74% ± 29%total cohort, 71% ± 32% R‑BPD and 77% ± 27% P‑BPD). This discrepancy highlights that coarse measures may overestimate adherence. Other adherence metrics in the literature include months with enough recordings for a management decision to be made [24], dichotomization into adhering and non-adherent [25] or categorizing by total number of entries [20].

In addition to health state evaluation and to discern potential impacts on BP, we examined BP values. The BP was within the range recommended for stroke patients in the total cohort and both groups [28]. Consistent with the even distribution of clinical characteristics, BP did not statistically differ between study groups. Imprecise BPD entries may cause misclassification of the severity of arterial hypertension and impact therapeutic decisions [26]. Hence, precision of BP documentation is an important determinant of the efficacy of BPDs [24]. Precision of documentation is measured by the proportion of entries ending in “0” or “5”, with expected proportions of one fifth in the absence of rounding [12]. The proportions observed in our cohort and in both study groups barely exceeded this threshold, with no statistically significant difference detected between study groups. We concluded that participants in both groups accurately documented their BP. The high baseline accuracy likely limited the opportunity to detect further improvements with P‑BPDs.

Individualized interventions empower patients, increasing their sense of control and potentially improving adherence [29]. Building on patient empowerment inherent to BPDs, we implemented personalization in the form of patient-provided images, additionally leveraging the emotional and associative impact of visuals. Consistently, we assessed the intervention’s effect on mental health. Participants with P‑BPD showed a 3-point lower score on the HADS depression scale on average but this difference did not reach statistical significance. We also observed statistically non-significant, mildly increased role limitations due to emotional problems and slightly higher anxiety in the intervention group. Given the small sample size and pilot nature of the study, these seemingly paradoxical findings require further exploration in a follow-up study.

Almost all study participants were satisfied with the use of BPDs and more than half shared their BPD at least once with other people, highlighting the emotional and associative benefits of BPDs; however, the insertion of pictures into BPDs did not increase the frequency of sharing or discussing the BPDs with other people. Those in the P‑BPD group submitted pictures showing family, friends, pets and holiday memories. They reported that the inserted images did increase their motivation to monitor their BP, extending the supportive effects of BPDs. While this did not translate into a significant difference regarding BPD adherence, patients in the control group felt that personalized images would improve their adherence. Most participants in both groups reported that they would prefer P‑BPDs over R‑BPDs in the future. Therefore, we regard our proposition of introducing personalized images to BPD to be well-received in our study.

As we conceived this investigation as a pilot study, we must report significant limitations evolving around our short observation period and small sample size. The rather short duration of our study limited our ability to assess whether adherence is sustained over a longer period. The short study period may explain why mean BP values as assessed in our study were a result of higher adherence to hypertension treatment following hospital discharge. Furthermore, we are unable to determine long-term effects of our method on BPD adherence. A longer observation period may have revealed different aspects of personalized BPDs and their suspected role in greater adherence. Furthermore, the second limitation due to the preliminary nature of our study was a small sample size. This has prevented us from detecting statistical differences and perform subgroup or regression analyses. A greater statistical power may have been helpful to corroborate our findings of certain patient reported outcomes (such as mental health aspects).

Also, age and sex matching might have been helpful. We will address these issues in a follow-up study for which we will perform power calculations based on our findings from this pilot study; however, we feel that we gained enough insight into patient preferences that enable the design of a methodologically sound follow-up study. Another limitation pertains to the limited interpretability of the findings due to discrepancies in the results. While most participants favored picture-personalized BPDs and reported increased motivation, this did not translate into significantly higher adherence. This suggests that visual personalization may enhance participant engagement but it does not necessarily lead to sustained behavioral change over time. Additionally, mood outcomes were inconsistent, with lower depression scores but slightly higher anxiety levels observed in the P‑BPD group. Although neither of these differences reached statistical significance, they underline the complex psychological effects of personalization. Lastly, we acknowledge that our study did not cover all aspects of BPD adherence, for example overreporting and underreporting of measured values [12].

In line with our intention to increase BPD adherence by fostering patient empowerment based on the inclusion of personalized images, we thoroughly explored how patients engaged with their BPDs. This strength of our study allowed us to evaluate factors affecting adherence other than inserted images (such as the presence of external reminders or previous experience with BPDs). Our open and real-world recruitment setting aiming to somewhat increase representability may be considered as another strength of our report.

In conclusion, one of the key findings of our pilot study was that the inclusion of personalized images to BPDs for people with stroke diagnoses is feasible. As BP management (e.g., through use of BPDs) remains a hallmark in secondary stroke prevention, any measure sustaining adherence over a long period is worth investigating. Our pilot study lays the ground for future research as we address relevant aspects of adherence to BP home monitoring and explored potential implications of patient preferences and perspectives. Lastly, our study enables robust sample size calculations and may facilitate the definition of other reasonable study endpoints in the future. These may include, besides BPD adherence, secondary adherence metrics such as precision of documentation, clinical impact of (P-)BPD use and psychological as well as behavioral outcomes such as patient engagement and mental health outcomes.