Please critique the research article (that is included here, pdf included here) from the following angles: a) Data collection b) Data analysis c) Results, findings and conclusion. 1. A minimum of 500

J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 1 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX (J Med Internet Res 2017;19(4):e125) doi: 10.2196/jmir .5604 KEYW ORDS smartphone; cell phone; Internet; mobile applications; health promotion; health beha vior Introduction As of 2015, nearly tw o-thirds (64%) of the American public owned a mobile phone, which is an increase from 35% in 2011 [1]. It is estimated that 90% of the w orldwide population will own a mobile phone by 2020 [ 1]. Current UK data re veals that mobile phone usage is increasing as 66% adults aged more than 18 years o wned a mobile phone in 2015, up from 61% in 2014 [2]. Mobile phone o wnership is higher among younger people, with 77% o wnership for those aged 16-24 years [ 3]. Although mobile phone o wnership is especially high among younger persons and those with higher educational attainment and income [ 4], those with lo wer income and educational attainment are no w lik ely to be “mobile phone dependent, ” meaning that the y do not ha ve broadband access at home and ha ve fe w other options for Web-based access other than via mobile phone. As mobile phone o wnership rapidly proliferates, so does the number of mobile phone softw are apps gro wn in the mark etplace [5]. Apps focused on health promotion are quite common: more than 100,000 health apps are a vailable in the iT unes and Google Play stores [ 6]. This staggering number speaks to both the huge mark et and ongoing demand for ne w tools to help the public manage their diet, f itness, and weight-related goals, and the limitations of the current health care system to pro vide such resources. A recent study found that 53% of cell phone users owned a smartphone—this translates to 45% of all American adults—and that half of those (or about 1 in 4 Americans) ha ve used their phone to look up health information [ 7]. There is increasing usage of health apps among health care professionals, patients and general public [ 8], and apps can play a role in patient education, disease self-management, remote monitoring of patients, and collection of dietary data [ 9-12]. Using mobile phones and apps, social media also can be easily accessed, and increasing numbers of indi viduals are using social media for health information with reported benef its and limitations [ 8]. Despite the massi ve uptak e in mobile phone o wnership and health app usage and their potential for impro ving health, important limitations of health apps are the lack of e vidence of clinical ef fecti veness, lack of inte gration with the health care deli very system, the need for formal e valuation and re vie w, and potential threats to safety and pri vacy [ 6,13-17]. Although pre vious studies ha ve described the sociodemographic f actors associated with mobile health and app use [ 7,18,19], it is a rapidly changing f ield with the most recent published reports reflecting data at least four to f ive years old. Additionally , there is a lack of information on the users of health apps in terms of their sociodemographic and health characteristics and health beha viors. Furthermore, to our kno wledge, there ha ve been no pre vious publications reporting on the association between the use of health apps, beha vioral or attitudinal f actors (ie, readiness or intentions to change), and health outcomes. This information is important for future health-impro ving initiati ves and for identifying appropriate use of health apps among population groups.

Therefore, the aim for our study w as 3-fold: (1) to describe the sociodemographic characteristics associated with health app use in a recent US nationally representati ve sample; (2) to assess the attitudinal and beha vioral predictors of the use of health apps for health promotion; and (3) to e xamine the association between the use of health-related apps and meeting the recommended guidelines for fruit and v egetable intak e and physical acti vity . Gi ven the increasing focus on ne w models for inte grating technology into health care and the need to expand the e vidence base on the role of health apps for health and wellness promotion, these research questions are timely and rele vant to inform the de velopment of health app interv entions. Methods Data Sour ce The National Cancer Institute’ s Health Information National Trends Surv ey (HINTS) is a national probability sample of US adults that assesses usage and trends in health information access and understanding. HINTS w as f irst administered in 2002-2003 as a cross-sectional surv ey of US ci vilians and noninstitutionalized adults. It has since been iterati vely administered in 2003, 2005, 2008, 2011, 2012, 2013, and 2014.

We used data from HINTS 4 Cycle 4 data released in June 2015, which corresponded to surv eys administered in August-No vember , 2014. Publicly a vailable datasets and information about methodology are a vailable at the HINTS website [ 20]. The 2014 iteration reported herein contained questions about whether participants used mobile phone or tablet technology and softw are apps for health-related reasons. The overall response rate w as 34.44%. This study w as re vie wed and qualif ied for an Ex emption by the American Academy of F amily Ph ysicians Institutional Re vie w Board. Participants A total of 3677 indi viduals completed the 2014 HINTS surv ey. From this sample, 148 respondents were considered partial completers, in that the y completed 50%-79% of the questions in Sections A and B. We included all 3677 respondents in our analysis. We used sampling weights from the HINTS dataset that were incorporated into the re gression analyses. Measur es Demographics We used participants’ self-report of their age, se x, race, ethnicity , income, le vel of education, English prof icienc y, height, and weight. We con verted height and weight into body mass inde x (BMI), using weight (kg)/height (m 2)×10,000, and classif ied participants as obese ( ≥30), o verweight (29.9-26), or normal weight or underweight (<26). J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 2 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX Usage of Mobile Devices and Health Apps We used participants’ responses to the 3 questions to characterize the distrib ution of subjects who used health-related softw are apps on their mobile de vices. The participants were ask ed whether the y had a tablet computer , smartphone, basic cell phone only , or none of the abo ve. We e xamined f actors for those with and without mobile de vices, since pre vious studies have sho wn dif ferences in seeking health information on the Internet related to access (e g, a vailability of a computer) [ 21,22], HINTS dataset is a nationally representati ve sample, and we wished to put our f indings on app use in the lar ger population conte xt. We cate gorized participants who had a mobile phone or a tablet de vice under the label “De vice+. ” Similarly , participants who did not report ha ving a mobile phone or a tablet device were labeled “De vice-. ” Of the De vice+ group, we also cate gorized them according to whether the y had health apps on their de vice (De vice+/App+) or did not ha ve health apps on their de vice (De vice+/App-). Fruit and Vegetable Intak e We assessed fruit and v egetable intak e using the 2 questions: amount of fruit consumed per day and amount of v egetables consumed per day (7 response options for each ranging from none to >4 cups per day). We reclassif ied the response options for both questions into a single dichotomous outcome v ariable, that is, the subject either (1) meets recommendations for fruit or v egetables (4 or more cups for each) or (2) does not meet recommendations for fruit or v egetables (all other response options). Fruit and v egetable scores were analyzed separately . Physical Activity We assessed ph ysical acti vity using the 2 questions: (1) in a typical week ho w man y days do you do an y ph ysical acti vity or e xercise of at least moderate intensity , such as brisk w alking, bic ycling at a re gular pace, and swimming at a re gular pace? (8 response options ranging from none to 7 days per week) and (2) on the days that you do an y ph ysical acti vity or e xercise of at least moderate intensity ho w long do you do these acti vities? (2 response options for minutes and hours). We reclassif ied the response options into a single dichotomous outcome v ariable for ph ysical acti vity , that is, whether the subject (1) met ph ysical acti vity recommendations ( ≥150 minutes per week) or did not meet the ph ysical acti vity recommendations (<150 minutes per week).

Intentions to Change Behavior We e xamined participants’ intentions to change beha vior based on the 5 questions (all with yes or no responses): At an y time in the last year , ha ve you intentionally tried to (1) increase the amount of fruit or 100% fruit juice you eat or drink, (2) increase the amount of v egetables or 100% v egetable juice you eat or drink, (3) decrease the amount of re gular soda or pop you usually drink in a week, (4) lose weight, and (5) increase the amount of e xercise you get in a typical week? Statistical Analysis The outcome v ariable (OUTCOME) w as a composite deri ved from 3 surv ey v ariables: (1) own a smartphone (an Internet-enabled mobile phone “suc h as iPhone andr oid Blac kBerry or Windows phone” dif fer entiated fr om a “basic cell phone ,” her eafter r eferr ed to as “mobile phone”) or de vice , (2) have health apps on mobile phone or de vice , and (3) use of health apps . Own a mobile phone or de vice w as a system-supplied deri ved v ariable to cate gorize responses gi ven to question B4 (possession of a mobile phone or tablet de vice). Have health apps on mobile phone or de vice (question B5) ask ed about health apps on a tablet or mobile phone. Use of health apps (question B6a) ask ed whether the apps on a mobile phone or tablet helped in achie ving a health-related goal. OUTCOME consisted of 3 le vels: De vice-/App- (33.2% of respondents), De vice+/App- (44% of respondents), and De vice+/App+ (22.77% of respondents). De vice referred to having a tablet or mobile phone, and App referred to ha ving a health-related app that ran on a tablet or mobile phone. A total of 93 of 3677 respondents were unable to be classif ied due to missing data. These people were not used in the analyses. To assess the relationship between OUTCOME and the demographic or health beha vior v ariables, simple unweighted 2-w ay crosstab tables were generated and tested with a chi-square test of association. We used a cutof f of P<.05 to determine statistical signif icance for all analyses. We used the R programming language (R-Studio) and SPSS (SPSS Inc) for all data modeling and analysis carried out in this study . Results Principal Findings From the 3677 total HINTS respondents, 3584 answered questions about whether or not the y had a tablet computer or mobile phone, or used apps. Figure 1 sho ws the participants in this study . J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 3 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX Figur e 1. Health Information National Trends Surv ey (HINTS) respondents’ use of mobile phones, tablets, and apps. Demographic Variables Associated W ith App Use Table 1 compares respondents grouped into De vice+/App+, De vice+/App-, and De vice-, according to sociodemographic characteristics. As sho wn in Table 1 , those who used health apps (compared with those who either did not ha ve apps or did not ha ve the necessary equipment) were more lik ely to be younger , li ve in metropolitan areas, ha ve more education, ha ve higher income, speak English well, be Asian, and report excellent health. There w as no signif icant association between both BMI and smoking status and app use. J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 4 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX Table 1. Demographic v ariables associated with app usage. P value Device- n (%) Device+/App- n (%) Device+/App+ nb,c (%) d Demographic v ariables .39 1156 (55.29) 1555 (50.23) 808 (51.62) Sex (female vs male; n a,c=3519) <.001 1111 (21.92) 1552 (52.25) 782 (65.62) Age (18-44 years vs 45+ years; n=3415) <.01 1121 (51.82) 1535 (27.95) 788 (12.72) Education (high school or less vs some colle ge or colle ge graduate, n=3444) <.001 1162 (75.12) 1560 (42.20) 808 (31.72) Income (US $0-49,999 vs 50,000 or greater; n=3530) <.01 1057 (83.68) 1453 (78.52) 763 (71.85) Race or ethnicity (white vs other; n=3273) .49 1114 (33.82) 1524 (36.98) 782 (33.71) BMI (normal vs o verweight, obese; n=3420) <.001 1191 (78.93) 1577 (85.67) 816 (92.10) Metro vs nonmetro (n=3584) <.001 1089 (90.37) 1497 (97.13) 759 (99.37) Speak English (v ery well or well vs not well or not at all; n=3584) <.001 1138 (74.99) 1544 (89.74) 795 (92.85) Self-rated health (e xcellent, v ery good, good vs f air or poor; n=3477) aThe sample sizes (n’ s) listed for each v ariable in the f ar left column represent the total number of respondents across all app-usage cate gories (De vice+/App+, De vice +/App-, De vice-) who answered that question. bThe sample sizes (n’ s) listed for each v ariable within each cell represent the total number of respondents within a gi ven app-usage cate gory (either Device+/App+, De vice +/App-, or De vice-) who answered that question. cSample sizes v ary for each v ariable due to missing v alues. dPopulation estimates were used for the numerators and denominators in the calculation of percentages. Ro w percentages do not add to 100%, as the table sho ws percentages within a gi ven app-usage cate gory (De vice+/App+, De vice +/App-, or De vice-). Association Between the Use of Apps and Intentions to Change Diet, P erf orm Ph ysical Acti vity , and Lose W eight Table 2 sho ws the association between the use of apps (v ersus De vice+/App- or De vice-) with intentions to change diet, perform ph ysical acti vity , or lose weight. As Table 2 sho ws, participants with apps were signif icantly more lik ely to report intentions to impro ve fruit ( P=.01) and v egetable consumption (P<.01), ph ysical acti vity ( P<.01), and weight loss ( P<.01) compared with those in the De vice+/App- or De vice- groups. Table 2. Association between the usage of apps for health-related goal and intentions to change diet, ph ysical acti vity , or lose weight. P value a Device- n (%) Device+/App- n (%) Device+/App+ n (%) Health-related intention .01 654 (48.94) 885 (58.50) 545 (63.76) Increase fruit <.01 717 (50.02) 1023 (64.26) 621 (74.92) Increase v egetables .06 754 (77.36) 1135 (82.76) 630 (84.96) Decrease soda <.01 769 (49.94) 1237 (65.42) 707 (82.99) Increase ph ysical acti vity <.01 881 (60.02) 1259 (71.75) 692 (83.36) Lose weight aSignif icance between participants with apps (De vice+/App+) compared with those not using apps or de vices (De vice+/App- or De vice- groups). Association Between the Use of Apps and Meeting Recommendations f or Fruit and Vegetable Intak e and Ph ysical Acti vity Table 3 sho ws the association between the use of apps (v ersus De vice+/App- or De vice-) and meeting the recommendations for fruit and v egetable intak e and ph ysical acti vity . P articipants in the De vice+/App+ group were not signif icantly more lik ely to meet recommendations for fruit and v egetables compared with those in the De vice+/App- or De vice- groups; ho we ver, the y were signif icantly more lik ely to e xercise more than 2 hours per week. J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 5 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX Table 3. Association between the use of apps for health-related goal and meeting recommendations for fruit and v egetables and ph ysical acti vity . P value a Device- n (%) Device+/App- n (%) Device+/App+ n (%) Percent respondents meeting recommendations .25 1161 (5.43) 1560 (7.96) 804 (8.87) Fruit .27 1155 (3.48) 1557 (3.01) 809 (4.81) Vegetables <.01 1144 (37.69) 1552 (47.79) 801 (56.23) Physical acti vity aSignif icance between participants with apps (De vice+/App+) compared with those not using apps or de vices (De vice+/App- or De vice- groups). Pr edicting Health App Adoption Only (Binary Classif ication) Table 4 presents the statistically signif icant odds ratios (ORs) as deri ved using multi variate logistic re gression when applied to the entire dataset. As e xpected, those aged 45-64 years (OR 0.56) or 65+ years (OR 0.19) had a reduced lik elihood of ha ving adopted health apps relati ve to younger persons. It also sho wed that males were slightly less lik ely (OR 0.80) to ha ve a health app compared with females. The most signif icant f inding w as the conf irmation that graduates had signif icantly higher odds (OR 2.83) of ha ving a health app especially when compared with those who had attained an education that w as considered “less than high school” (OR 0.43). The results also indicated that the cate gory “completed high school only” had no predicti ve ability for estimating whether a person had adopted a health app. Table 4. Statistically signif icant odds ratios deri ved using multi variate logistic re gression when applied to the entire dataset for predicting health app adoption only . P value Odds ratio (95% CI) Variable <.001 0.56 (0.47-0.68) Age (45-64 years) <.001 0.19 (0.14-0.24) Age (65+ years) <.01 0.80 (0.66-0.94) Sex (male) <.001 2.83 (2.18-3.70) Education (colle ge graduate or higher) <.01 0.43 (0.24-0.72) Education (less than high school) <.01 1.70 (1.30-2.26) Education (some colle ge) .05 1.25 (0.99-1.55) Race (black) Pr edicting Mobile Technology Adoption Only (Binary Classif ication) Table 5 presents the statistically signif icant ORs that increased or decreased the lik elihood that a person had adopted mobile technology (tablet or mobile phone). Interestingly , there were no statistically signif icant ORs for gender or racial cate gories. Ho we ver, similar to predicting health app adoption, both age and education were signif icant v ariables for predicting whether a person had adopted a mobile de vice, especially if that person was a colle ge graduate (OR 3.30). In addition, the results indicated that the cate gory “completed high school only” had no predicti ve ability for estimating whether a person had adopted a mobile de vice. J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 6 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX Table 5. Statistically signif icant odds ratios deri ved using multi variate logistic re gression when applied to the entire dataset for predicting mobile de vice adoption only . P value Odds ratio (95% CI) Variable <.001 0.35 (0.28-0.45) Age (45-64 years) <.001 0.09 (0.07-0.12) Age (65+ years) <.001 3.30 (2.65-4.11) Education (colle ge graduate or higher) <.001 0.51 (0.37-0.70) Education (less than high school) <.001 1.87 (1.50-2.32) Education (some colle ge) Discussion Principal Findings Our f irst objecti ve w as to describe the sociodemographic and health beha vior characteristics associated with health app use in a recent US nationally representati ve sample. Consistent with pre vious f indings [ 7], we found that those who were younger , had more education, reported e xcellent health, and had a higher income were more lik ely to use health apps. Our predicti ve modeling using multi variate logistic re gression sho wed that education, se x, gender , and race were only mildly to moderately potent in predicting mobile technology adoption.

Our second objecti ve w as to assess the beha vioral and attitudinal predictors of the use of health apps for health promotion. We found that participants with apps were also more lik ely to report intentions to impro ve fruit and v egetable consumption, ph ysical acti vity , and weight loss. Finally , the third objecti ve w as to examine the association between the use of health-related apps and meeting the recommended guidelines for fruit and v egetable intak e and ph ysical acti vity . We found that participants in the health apps group were signif icantly more lik ely to meet recommendations for ph ysical acti vity compared with those without a de vice or health apps. Comparison W ith Prior W ork This study shares some similarities with pre vious HINTS analyses. F or e xample, McCully et al [ 19] reported that users of the Internet for diet, weight, and ph ysical acti vity tended to be younger and more educated and that Internet use for these purposes w as more lik ely to be associated with higher fruit and vegetable intak e and moderate e xercise. Ho we ver in that study , women were no more lik ely than men to use the Internet for diet, weight, and ph ysical acti vity , which w as dif ferent from our f indings. In that study , minorities were more lik ely to use the Internet; in our study , we found no such association. Consistent with our f indings, K ontos et al found that males, those with lo wer education, and older US adults were less lik ely to eng age in a number of eHealth acti vities [ 18]. Similar to their findings 3 years ago, our f indings pointed to dif ferences by education for app use for health promotion.

The association between app use, intention to change lifestyle beha viors, and actually meeting recommendations for health y lifestyle f actors is interesting and could be due to se veral reasons. First, it is possible that there are pree xisting dif ferences in indi viduals who eng age with health apps compared with those who do not. Users of health apps may ha ve greater moti vation and interest in changing their diet, weight, or ph ysical acti vity . A recent re vie w found that v ery fe w a vailable apps pro vided evidence-based support to meet lifestyle recommendations [ 13]. It could also be that app users are eng aging with health apps to help them simply track or self-manage dif ferently than their counterparts; thus, there could be dif ferences in preferences or needs. Due to the correlational nature of the data, we cannot dra w conclusions about the relationships or causal pathw ays. Similar observ ations ha ve been reported in a study of users of the Internet for diet, weight, and ph ysical acti vity promotion [19]. The pre valence of app usage in our study w as 22% (816/3677). This is a doubling from the K ontos study in which 11.7% downloaded info onto a mobile de vice. Although the questions in these 2 HINTS datasets were w orded dif ferently (e g, “do wnloaded” is broader and not referring e xclusi vely to downloading an app), it suggests that demand for apps continues to rise and of fers potential for reaching a gro wing se gment of the US population.

Our f indings pro vide e vidence for educational, age, and gender dif ferences in the use of mobile de vices and health apps. Educational attainment, age, and gender ha ve been pre viously sho wn to be important predictors of adoption of mobile de vices and apps [ 18]. Educational attainment appears more important than other v ariables commonly used as proxies for socioeconomic position (e g, income, race or ethnicity). The reasons for the educational dif ferences are unclear , b ut may reflect skills and conf idence with the use of de vices and possibly social norms related to percei ved v alue. Similarly , age lik ely reflects both social norms and cohort ef fects, that is, e xposure during younger ages to these de vices and apps. The reasons for gender dif ferences are less clear , but may reflect dif ferences in health-seeking beha vior , and interest and participation in health y lifestyle interv entions generally . Limitations This study had limitations that should be k ept in mind when interpreting results. First, HINTS is a cross-sectional surv ey; although it is a nationally representati ve cohort of indi viduals, we were not able to e valuate the trends in an indi vidual’ s health app use o ver time. There is the possibility of unmeasured confounding, that is, unidentif ied f actors that might be associated with app use and intentions or health beha viors, which could influence the interpretation of results. Although the results sho wed association, it did not indicate a causal relationship. This study could not answer the question of whether more moti vated indi viduals sought out apps, or whether J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 7 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX app use impro ved moti vation and health outcomes. Furthermore, some of the cells for subgroups were small, thereby limiting the generalizability of some of the subanalyses. As with all cross-sectional surv eys, this w as a study of association, not causation. Finally , we were limited by the questions that were ask ed in the HINTS surv ey. For e xample, we did not ha ve details about specif ic health apps or features of apps used, the intensity of use, whether the apps were interacti ve and link ed to other health promotion supports (e g, telehealth), and other strate gies used for health beha vior change. Despite these limitations, the results did identify areas for future research and add to the kno wledge base about predictors of the use of health apps. Conclusions Compared with pre vious studies, man y indi vidual sociodemographic f actors are becoming less important in influencing eng agement with mobile de vices and health app use; ho we ver, dif ferences persist for gender , age, and educational attainment. As health care under goes technological transformation with its electronic health records systems and indi viduals’ access to their records, there are man y opportunities for clinical care models to be e xpanded and impro ved, perhaps through the use of apps as a means for sharing data, although this remains an unanswered question. This study contrib utes to the literature by pro viding up-to-date information on populations most and least lik ely to use health apps to guide clinical interv entions, commercial de velopers, and public health programs when designing eHealth technology . Conflicts of Inter est None declared.

Refer ences 1. Ericsson. 2015. Ericsson Mobility Report: On the pulse of the netw ork ed society URL: http://www .ericsson.com/res/docs/ 2015/ericsson-mobility-report-june-2015.pdf [accessed 2016-01-22] [ WebCite Cache ID 6ejSFiicz ] 2. OfCom. 2015. Smartphone usage URL: http://media.ofcom.or g.uk/f acts/ [accessed 2016-01-22] [ WebCite Cache ID 6ejSSqrN7 ] 3. OfCom. Belf ast: OfCom; 2014. Telecommunications f acts and f igures URL: https://www .ofcom.or g.uk/ [accessed 2017-02-27] [WebCite Cache ID 6ob5yNLrU ] 4. Smith A. Pe w Research Center . 2015. The Smartphone Dif ference URL: http://www .pe winternet.or g/2015/04/01/ us-smartphone-use-in-2015/ [accessed 2017-02-28] [ WebCite Cache ID 6ejS9bn6M ] 5. Boudreaux ED, Waring ME, Hayes RB, Sadasi vam RS, Mullen S, P agoto S. Ev aluating and selecting mobile health apps: strate gies for healthcare pro viders and healthcare or ganizations. Transl Beha v Med 2014 Dec;4(4):363-371 [ FREE Full text] [doi: 10.1007/s13142-014-0293-9 ] [Medline: 25584085 ] 6. Research2guidance. Research2guidance URL: http://research2guidance.com [accessed 2016-01-22] [ WebCite Cache ID 6ejS4CO9X ] 7. Fox S, Dugg an M. Pe w Research Center . Mobile Health 2012: Half of smartphone o wners use their de vices to get health informationone-f ifth of smartphone o wners ha ve health apps URL: http://www .pe winternet.or g/2012/11/08/ mobile-health-2012/ [accessed 2017-02-27] [ WebCite Cache ID 6ob6C9mfG ] 8. Moorhead SA, Hazlett DE, Harrison L, Carroll JK, Irwin A, Ho ving C. A ne w dimension of health care: systematic re vie w of the uses, benef its, and limitations of social media for health communication. J Med Internet Res 2013;15(4):e85 [ FREE Full te xt] [doi: 10.2196/jmir .1933 ] [Medline: 23615206 ] 9. Zhu F , Bosch M, Woo I, Kim S, Boushe y CJ, Ebert DS, et al. The use of mobile de vices in aiding dietary assessment and evaluation. IEEE J Sel Top Signal Process 2010 Aug;4(4):756-766 [ FREE Full te xt] [doi: 10.1109/JSTSP .2010.2051471 ] [Medline: 20862266 ] 10. Mosa AS, Yoo I, Sheets L. A systematic re vie w of healthcare applications for smartphones. BMC Med Inform Decis Mak 2012;12:67 [ FREE Full te xt] [doi: 10.1186/1472-6947-12-67 ] [Medline: 22781312 ] 11. O'Malle y G, Do wdall G, Burls A, Perry IJ, Curran N. Exploring the usability of a mobile app for adolescent obesity management. JMIR Mhealth Uhealth 2014;2(2):e29 [ FREE Full te xt] [doi: 10.2196/mhealth.3262 ] [Medline: 25098237 ] 12. O'Malle y G, Clark e M, Burls A, Murph y S, Murph y N, Perry IJ. A smartphone interv ention for adolescent obesity: study protocol for a randomised controlled non-inferiority trial. Trials 2014;15:43 [ FREE Full te xt] [doi: 10.1186/1745-6215-15-43 ] [Medline: 24485327 ] 13. Knight E, Stuck ey MI, Prapa vessis H, Petrella RJ. Public health guidelines for ph ysical acti vity: is there an app for that? A re vie w of android and apple app stores. JMIR Mhealth Uhealth 2015;3(2):e43 [ FREE Full te xt] [doi: 10.2196/mhealth.4003 ] [Medline: 25998158 ] 14. Pagoto S, Schneider K, Jojic M, DeBiasse M, Mann D. Evidence-based strate gies in weight-loss mobile apps. Am J Pre v Med 2013 No v;45(5):576-582. [doi: 10.1016/j.amepre.2013.04.025 ] [Medline: 24139770 ] 15. Breton ER, Fuemmeler BF , Abroms LC. Weight loss-there is an app for that! But does it adhere to e vidence-informed practices? Transl Beha v Med 2011 Dec;1(4):523-529 [ FREE Full te xt] [doi: 10.1007/s13142-011-0076-5 ] [Medline: 24073074 ] J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 8 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX 16. Abroms LC, P admanabhan N, Tha weethai L, Phillips T. iPhone apps for smoking cessation: a content analysis. Am J Pre v Med 2011 Mar;40(3):279-285 [ FREE Full te xt] [doi: 10.1016/j.amepre.2010.10.032 ] [Medline: 21335258 ] 17. Eng DS, Lee JM. The promise and peril of mobile health applications for diabetes and endocrinology . Pediatr Diabetes 2013 Jun;14(4):231-238 [ FREE Full te xt] [doi: 10.1111/pedi.12034 ] [Medline: 23627878 ] 18. Kontos E, Blak e KD, Chou WS, Prestin A. Predictors of eHealth usage: insights on the digital di vide from the Health Information National Trends Surv ey 2012. J Med Internet Res 2014;16(7):e172 [ FREE Full te xt] [doi: 10.2196/jmir .3117 ] [Medline: 25048379 ] 19. McCully SN, Don BP , Upde graf f JA. Using the Internet to help with diet, weight, and ph ysical acti vity: results from the Health Information National Trends Surv ey (HINTS). J Med Internet Res 2013;15(8):e148 [ FREE Full te xt] [doi: 10.2196/jmir .2612 ] [Medline: 23906945 ] 20. HINTS. 2015. Health Information National Trends Surv ey URL: http://hints.cancer .go v/ [accessed 2016-01-22] [ WebCite Cache ID 6ejRyxj9Y ] 21. Finne y RL, Hesse BW , Moser RP , Ortiz MA, K ornfeld J, Vanderpool RC, et al. Socioeconomic and geographic disparities in health information seeking and Internet use in Puerto Rico. J Med Internet Res 2012 Jul;14(4):e104 [ FREE Full te xt] [doi: 10.2196/jmir .2007 ] [Medline: 22849971 ] 22. Kontos EZ, Bennett GG, Visw anath K. Barriers and f acilitators to home computer and internet use among urban no vice computer users of lo w socioeconomic position. J Med Internet Res 2007;9(4):e31 [ FREE Full te xt] [doi: 10.2196/jmir .9.4.e31 ] [Medline: 17951215 ] Ab br eviations HINTS: Health Information National Trends Surv ey Edited by G Eysenbac h; submitted 04.02.16; peer -reviewed by J Upde graff, A Burls, B Fuemmeler; comments to author 06.04.16; revised ver sion r eceived 18.05.16; accepted 21.06.16; published 19.04.17 Please cite as:

Carr oll JK, Moorhead A, Bond R, LeBlanc WG, P etrella RJ , Fiscella K Who Uses Mobile Phone Health Apps and Does Use Matter? A Secondary Data Analytics Appr oac h J Med Internet Res 2017;19(4):e125 URL: http://www .jmir .org/2017/4/e125/ doi: 10.2196/jmir .5604 PMID: 28428170 ©Jennifer K Carroll, Anne Moorhead, Raymond Bond, William G LeBlanc, Robert J Petrella, K evin Fiscella. Originally published in the Journal of Medical Internet Research (http://www .jmir .or g), 19.04.2017. This is an open-access article distrib uted under the terms of the Creati ve Commons Attrib ution License (http://creati vecommons.or g/licenses/by/2.0/), which permits unrestricted use, distrib ution, and reproduction in an y medium, pro vided the original w ork, f irst published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www .jmir .or g/, as well as this cop yright and license information must be included. J Med Internet Res 2017 | v ol. 19 | iss. 4 | e125 | p. 9 http://www .jmir .org/2017/4/e125/ (page number not for citation purposes) Carroll et al JOURN AL OF MEDICAL INTERNET RESEARCH XSL •FO RenderX