Which cancer patients access their clinical notes? A disparities analysis

Abstract

Purpose:

Access to clinical notes enhances patient engagement and trust, and the 21^st Century Cures Act enabled immediate electronic patient access in April 2021. Yet, technological advances may perpetuate disparities, which remain understudied. Understanding whether inequities in note access exist in oncology would highlight challenges around making this foundational health information available to all patients receiving ongoing, complex medical care.

Materials and Methods:

This study at a high-volume specialty cancer center explored disparities around clinical notes posted to patients’ portal accounts from 9/1/2021 to 8/31/2022 and accessed by 3/19/2024. Logistic and Poisson regression were used to identify patient characteristics associated with note access and note opening rates.

Results:

The study included 124,554 patients and 815,104 clinical notes, 43.7% of which (356,290) were accessed. While modest differences in access rates emerged around sex, age, and marital status, larger disparities appeared for ethnicity, race, and language: Black patients (OR=0.63, 95% CI: 0.60-0.66,; p<0.001; IRR=0.74, 95% CI: 0.73-0.76; p<0.001), Hispanic patients (OR=0.85, 95% CI: 0.80-0.90; p<0.001; IRR=0.90, 95% CI: 0.89-0.92; p<0.001), and non-English preferred language speakers (OR=0.78, 95% CI: 0.72-0.84; p<0.001; IRR=0.82, 95% CI: 0.80-0.84; p<0.001) were, respectively, 37%, 15%, and 22% less likely to open at least one note, and opened 26%, 10%, and 18% percent fewer notes, compared to white, non-Hispanic, and English-preferred patients, respectively.

Conclusion:

This analysis highlighted several disparities, most notably by race, ethnicity, and language, in cancer patients’ accessing clinical notes. Tailored interventions are crucial to ensure that diverse groups benefit from digital healthcare resources.

PURPOSE

Clinical notes are a vital part of the medical record, documenting the provider’s impressions and the patient’s care plan. Patients with access to their notes report being more satisfied with care, having more trust in their providers, and being more engaged with care[1-7]. Some evidence indicates, as well, that adherence to prescribed medication regimens may be higher with access to notes[8, 9].

Until recently, however, patients lacked immediate, ubiquitous access to clinical notes. In 1996, when HIPAA established patients’ right to access their health information, provider organizations began making some clinical data – for example, laboratory results and imaging reports -- available electronically via patient portal [10]. However, the process for patients to access the complete medical record, including the clinical documentation, typically required a written request, which might be a protracted and burdensome process.

This situation changed dramatically with the 21^st Century Cures Act[11]. As of April 2021, these provisions ensured patients’ rapid electronic access to clinic notes, among other data. For patients today, electronic access to clinical notes is both ubiquitous and immediate.

To ensure that all patients benefit from this transformation, it is important to understand any disparities that might be created, perpetuated, or amplified as technology enables expanded electronic access to clinical data. To be sure, new technologies often are adopted unevenly across patient groups [12, 13]. Yet, while disparities in portal access are well-studied[12, 14-17], only a few studies have examined clinical note access by diverse patient groups. For example, disparities have been documented in patients accessing their emergency room [18] and primary care notes [19, 20]. In these studies, patients who were younger (though over 18), white, English-speaking, married [21, 22], and privately insured were more likely to access their notes. However, to our knowledge, no data address the extent to which disparities in accessing notes exist for cancer patients, who may have a different level of engagement with their care due to the seriousness of the diagnosis and ongoing, complex medical care. To our knowledge, this study is the first to focus on disparities in clinical note access among this critical population.

OBJECTIVE

While over 100 studies document disparities in patients’ portal access[14], the question of equitable access to clinical notes is a vital yet understudied topic. Understanding whether inequities in access to notes exist in oncology would help establish a more complete view of the challenges around providing patients with access to this foundational health information. This study investigated these crucial issues in the wake of the 21^st Century Cures Act: Using data from a high-volume cancer center in New York City, we assessed the extent to which characteristics of adult patients with active portal accounts were associated with differences in rates of accessing clinical notes. We hypothesized that disparities might exist across patient characteristics such as race, ethnicity, preferred language, marital status, age, and sex assigned at birth.

MATERIALS AND METHODS

This retrospective, observational study took place at Memorial Sloan Kettering Cancer Center (MSK), a high-volume cancer center with facilities in New York City and the surrounding metropolitan area. At the time of this study, MSK was using an internally developed patient portal, MyMSK, which notified patients daily when new results and documents were available. This study was determined by the MSK IRB to be exempt.

Data:

Data were obtained from MSK clinical and administrative data systems, including patients’ demographic information (age, sex, race, ethnicity, marital status, preferred language). The sample included all adult (aged 18+) MSK patients for whom at least one clinical note was created and available via the MyMSK portal during a one-year period (9/1/2021 to 8/31/2022). We considered a note “accessed” if it was opened by the patient or a proxy by 3/19/2024, the date of data retrieval. Since a proxy (e.g., a spouse or adult child) will often use the patient’s own account to access data in the portal rather than a designated proxy account [23], in this study, we could not accurately disaggregate access by a patient from access by a proxy on a patient’s behalf. Rather, we considered them together, broadly counting both as “patient access.”

Analysis plan:

Summary statistics, medians, and interquartile ranges (IQRs) were used to describe continuous variables, with counts and percents for categorical variables. Our primary goals were to assess patient access to notes, defined as if patients read any vs. no notes, and the rate of note opening. Since we saw a difference between patients who accessed versus never accessed notes, the analysis of rate of note opening only included patients who opened at least 1 note, to avoid over inflating notes that were not accessed. Univariable and multivariable logistic and Poisson regressions were used to assess the association between demographic factors and the two outcomes, respectively. An interaction between marital status and sex was considered in both analyses Backward selection was used for the multivariable analysis, starting with a set of candidate variables with p<0.05 in the univariable analysis. R version 4.3.2 was used for all analyses. All tests were two-sided, and p < 0.05 was considered significant.

RESULTS

The study period included a total of 815,104 clinical notes authored on 124,554 adult patients. The median age of patients was 63 years, and 59% were female (Table 1); 84% were white, 93.7% non-Hispanic, 70% partnered and 97% had English as a preferred language. In all, 92,036 (73.8%) patients accessed 356,290 (43.7%) notes. The median number of notes created per patient was 3 (IQR 1,8). The median number of notes accessed per patient was 1.0 IQR (0.0, 3.0). Among patients that accessed at least 1 note, the median number of notes created per patient was 4 (IQR: 2, 9) and number accessed per patient or proxy was 2.0 (IQR: 1.0, 4.0).

Table 1.

Patient Characteristics

	All N = 124,554	Accessed 0 notes N = 32,518	Accessed 1+ note N = 92,036
Age (years)	63 (52, 72)	62 (52, 71)	63 (53, 72)
Sex
F	73,960 (59%)	19,786 (61%)	54,174 (59%)
M	50,594 (41%)	12,732 (39%)	37,862 (41%)
Race
White	100,313 (84%)	25,286 (81%)	75,027 (85%)
Asian	8,899 (7.4%)	2,357 (7.6%)	6,542 (7.4%)
Black or African American	7,083 (5.9%)	2,500 (8.0%)	4,583 (5.2%)
Other	3,324 (2.8%)	1,033 (3.3%)	2,291 (2.6%)
Unknown	4,935	1,342	3,593
Ethnicity
Not Hispanic or Latino	110,423 (94%)	28,544 (92%)	81,879 (94%)
Hispanic or Latino	7,474 (6.3%)	2,336 (7.6%)	5,138 (5.9%)
Unknown	6,657	1,638	5,019
Marital Status
Partnered	87,086 (70%)	22,156 (68%)	64,930 (71%)
Single/Separated/Other	37,468 (30%)	10,362 (32%)	27,106 (29%)
Preferred Language
English	120,194 (97%)	31,147 (96%)	89,047 (97%)
Other	4,193 (3.4%)	1,311 (4.0%)	2,882 (3.1%)
Unknown	167	60	107

¹Median (IQR); n (%)

Patients accessing at least 1 note:

In univariable analysis (Table 2), all demographic variables were associated with the outcome of accessing at least 1 note. The multivariable model thus included sex, marital status, the interaction between sex and marital status, race, ethnicity, preferred language, and age (in 10-year increments), of which all were independently associated with accessing at least 1 note except Asian race as compared to white patients:

Table 2.

Odds of Accessing At least One Note by Patient Characteristics, Univariable and Multivariable Logistic Regression Models

Characteristic	UVA					MVA
	N	Event N	OR1	95% CI1	p-value	OR1	95% CI1	p-value
Sex
F	73,960	54174	—	—		—	—
M	50,594	37862	1.09	1.06, 1.11	<0.001	1.08	1.05, 1.12	<0.001
Ethnicity
NOT HISPANIC	110,423	81879	—	—		—	—
HISPANIC OR LATINO	7,474	5138	0.77	0.73, 0.81	<0.001	0.85	0.80, 0.90	<0.001
Race
White	100,313	75027	—	—		—	—
Asian	8,899	6542	0.94	0.89, 0.98	0.008	0.96	0.91, 1.01	0.2
Black/African American	7,083	4583	0.62	0.59, 0.65	<0.001	0.63	0.60, 0.66	<0.001
Other	3,324	2291	0.75	0.69, 0.81	<0.001	0.87	0.80, 0.95	0.001
Primary Language
English	120,194	89047	—	—		—	—
Other	4,193	2882	0.77	0.72, 0.82	<0.001	0.78	0.72, 0.84	<0.001
Marital Status
Partnered	87,086	64930	—	—		—	—
Single/Separated/Other	37,468	27106	0.89	0.87, 0.92	<0.001	0.97	0.94, 1.01	0.2
Age (10-year increments)	124,554	92036	1.04	1.03, 1.05	<0.001	1.03	1.02, 1.04	<0.001
Sex * Marital Status
M * Single/Separated/Other						0.91	0.85, 0.96	0.002

¹OR = Odds Ratio, CI = Confidence Interval

• The interaction between marital status and sex was significant (p=0.002).Partnered men were the most likely to open a note compared to single men and single or partnered women. There was no difference between single and partnered women (predicted probabilities: partnered males (0.695, 95% CI: 0.69-0.71), non-partnered males (0.668, 95% CI: 0.66-0.68), partnered females (0.678, 95% CI: 0.67-0.69), and non-partnered females (0.673, 95% CI: 0.66-0.68) (Figure 1).

• Ethnicity: Hispanic patients were 15% less likely to access at least 1 note compared to non-Hispanic patients (OR=0.85, 95% CI (0.80, 0.90); p<0.001).

• Race: Relative to white patients, patients who were Black and in “other” racial categories were 37% (OR=0.63, 95% CI (0.60, 0.66); p<0.001) and 13% (OR = 0.87, 95% CI (0.80, 0.95); p=0.001) less likely to access at least 1 note, respectively. Asian patients were not statistically different than white patients (OR=0.96, 95% CI (0.91, 1.01); p=0.2)

• Preferred language: Patients whose preferred language was not English were 22% less likely to access at least 1 note (OR=0.78, 95% CI (0.72, 0.84); p<0.001), relative to English-preferred patients.

• Age: With each 10-year increase in age, the likelihood of accessing at least 1 note increased by 3% (OR = 1.03, 95% CI (1.02, 1.04); p<0.001).

Figure 1: Odds of Accessing Notes for Interaction between Marital Status and Sex.

Rates of a patient accessing available notes:

In univariable analysis all demographic variables were associated with the outcome of note accessing rates (Table 3). In the multivariable model, sex, marital status, the interaction between sex and marital status, race, ethnicity, preferred language, and age (in 10-year increments) were independently associated with rates of note opening:

Table 3.

Note Access Rates^§ by Patient Characteristics, Univariable and Multivariable Poisson Regression Models^**

Characteristic	UVA				MVA
	N	IRR1	95% CI1	p-value	IRR1	95% CI1	p-value
Sex
F	54,174	—	—		—	—
M	37,862	0.98	0.97, 0.99	<0.001	0.97	0.96, 0.97	<0.001
Ethnicity
NOT HISPANIC	81,879	—	—		—	—
HISPANIC OR LATINO	5,138	0.85	0.84, 0.87	<0.001	0.90	0.89, 0.92	<0.001
Race
White	75,027	—	—		—	—
Asian	6,542	0.93	0.92, 0.95	<0.001	0.96	0.94, 0.97	<0.001
Black/African American	4,583	0.74	0.72, 0.75	<0.001	0.74	0.73, 0.76	<0.001
Other	2,291	0.83	0.81, 0.85	<0.001	0.91	0.88, 0.93	<0.001
Primary Language
English	89,047	—	—		—	—
Other	2,882	0.81	0.79, 0.83	<0.001	0.82	0.80, 0.84	<0.001
Marital Status
Partnered	64,930	—	—		—	—
Single/Separated/Other	27,106	0.93	0.92, 0.94	<0.001	0.95	0.94, 0.97	<0.001
Age (10-year increments)	92,036	1.01	1.01, 1.02	<0.001	1.01	1.01, 1.01	<0.001
Sex * Marital Status
M * Single/Separated/Other					0.97	0.95, 0.99	0.003

¹IRR = Incidence Rate Ratio, CI = Confidence Interval

^§Among patients opening at least 1 note.

^**Missing values exist for race, ethnicity and preferred language.

• The interaction between marital status and sex was significant (p=0.003): Regardless of gender, partnered patients had a higher note opening rate, but the difference in predicted rates between non-partnered vs. partnered patients was larger in males (1.621, 95% CI: 1.59-1.65 versus 1.748, 95% CI: 1.72-1.78) than females (1.728, 95% CI: 1.70-1.76 versus 1.810, 95% CI: 1.78-1.84). Partnered males had a similar rate to non-partnered females (Figure 2).

• Ethnicity: Hispanic patients accessed 10% fewer notes than non-Hispanic patients. (IRR = 0.90, 95% CI (0.89, 0.92); p<0.001).

• Race: Compared to white patients, patients self-identifying as Asian, Black, and other racial categories accessed 4% (IRR = 0.96, 95% CI (0.94, 0.97); p<0.001), 26% (IRR = 0.74, 95% CI (0.73, 0.76); p<0.001), and 9% (IRR = 0.91, 95% CI (0.88, 0.93); p<0.001) fewer notes.

• Preferred language: Patients whose preferred language was not English accessed 18% (IRR = 0.82, 95% CI (0.80, 0.84); p<0.001) fewer notes than English-preferred patients.

• Age: With each 10-year increase in age, patients accessed 1% more notes (IRR = 1.01, 95% CI (1.01, 1.01); p<0.001).

Figure 2: Note access rates for interaction between marital status and sex.

DISCUSSION

This study investigated patterns in the rates at which cancer patients with active portal accounts accessed their clinical notes once these became electronically available under the 21^st Century Cures Act. There is some literature about note access rates in general; however, to our knowledge this study is the first to examine disparities in note access rates in a cancer setting. Overall, we found that access rates were relatively high, with nearly 125,000 patients collectively accessing 43.7% of over 800,000 available notes. Among patients that accessed at least 1 note, this proportion was even higher, 51.2%. This access rate is consistent with that found at other organizations that notify patients when notes are available [20], as was the case in our study.

We found disparities of varying magnitudes across a variety of demographic characteristics, most notably race and ethnicity. Concurring with similar studies of clinical data access disparities [13, 18, 24, 25], our analysis revealed that Hispanic, Black, Asian, and other non-white patients were less likely to access any notes, and/or accessed fewer notes, than non-Hispanic and white patients – with the widest disparities among Hispanic and Black patients. Prior research shows Black and Asian patients reporting less positive interactions with providers than whites [24], which may help explain the lower rates by race and ethnicity in our study. Additional patient-related barriers may exist, due to technology aspects, such as functional and usability barriers that may disproportionally affect certain groups (e.g., older adults, patients who prefer non-English languages).

Research also indicates that “clinician encouragement” is an important factor in getting people to access their medical records online [25, 26], so variations in such encouragement could account for some of the racial and ethnic disparities we observed. Future work should address potential opportunities to increase clinical comfort with encouraging diverse patients to use the portal, such as developing scripts for providers to use with patients. Additionally, healthcare provider organizations – both oncologic and non-oncologic – should track use of patient-facing technology across the segments of the patients that they serve. Support for the use of patient technology should be made available to assure that all patients can benefit from the potential of technology and that disparities, where they are present, are ameliorated.

We also found that patients who preferred languages other than English had significantly lower odds of accessing at least 1 note, as well as lower rates of note access. While only 3% of our sample of portal users preferred a non-English language, this finding comports with prior research showing that English-preferred patients were more likely to access clinical documentation in non-oncologic contexts [18-20]. Drivers of this disparity may include not only physician encouragement but also technology access barriers, variable awareness of and interest in portals, and portal use by friends and family [27]. Further studies are needed to examine how both portal adoption and clinic note access can be enhanced among this vulnerable population.

Additionally, our analysis identified differences by sex and marital status, with notable interaction effects. Marital status appeared more meaningful among men than women: Partnered men were the most likely of any group to open any notes, and unpartnered men were least likely, with no differences between partnered and unpartnered women. Similarly, while partnered men and women both had higher note opening rates than unpartnered peers, the effect of marital status was greater among men than women. We can only speculate about the specific reasons for these complex findings – it may be, for example, that men are more likely than women to rely on proxies to help them stay abreast of their clinic notes during cancer treatment, or unpartnered men and women may have less support in general for health-related activities or less time to use digital healthcare tools. In any case, our findings are consistent with prior studies identifying lower healthcare engagement among male patients [28-34] and underscoring the critical role of caregivers in patients’ engagement with digital healthcare resources [35-37]. Ultimately, these disparities we observed by sex and marital status suggest the need for greater support for unpartnered patients. Further research is critical to developing targeted interventions that will promote engagement across a diverse spectrum of patients.

We found that older patients were marginally more likely to access any notes, and to access more notes, than younger patients. This finding is unexpected, as other studies have documented significantly lower technology usage among older patients[28-34, 38-40]. It may be that proxies are driving this increased access by older patients. For example, it could also be that, in cancer, proxies who are adult children may be more interested in following the course of their parent’s disease than in other clinical situations.

In general, however, the disparities we identified by age, marital status, and sex were relatively modest. More than addressing these disparities, a greater challenge may be ensuring equitable enrollment in the portal in the first place. Indeed, many studies have documented persistent and sizeable disparities -- by age, ethnicity, sex, and other demographics -- in patients’ adoption of new healthcare information technologies including but not limited to portal enrollment [26, 35-37, 41-45]. With the 21st Century Cures Act making the portal an increasingly valuable resource for patients, it is vital to ensure that all patients have equal opportunity to benefit from it.

It also seems appropriate for us to comment on the considerable challenges associated with proxy accounts. Initially we had hoped to investigate potential differences in note access rates between patients and proxies, as for many vulnerable groups – particularly patients who face technology or language barriers -- proxies can be vital lifelines. However, as described in prior research, it is also common for proxies to use the patient’s own account, rather than a designated proxy account, to access portal information [13, 23], so we were unable to determine accurately the extent to which proxies were accessing clinic notes. Others have noted the challenges with – and the importance of – understanding when a proxy, as opposed to the patient, is accessing a portal account. Further research and interventions are needed to ensure equitable information access among patients who rely on proxies.

Our findings should be viewed in light of several limitations. Our data come from a single cancer center in an urban setting so the generalizability of our results may be limited. Further, in assessing rates of note access, we were guided by an assumption that information access is good for patients. However, we did not examine the actual outcomes of patients’ accessing their notes, including the extent to which patients comprehended, appreciated, used, or otherwise benefited from the information therein. It is also possible that patients experienced negative effects from viewing their notes – for example, they may have felt anxiety, distress, or confusion. While assessing these outcomes was beyond the scope of this study, our findings underscore that a great many patients are interested in accessing their clinic notes -- and that multiple disparities mark the use of this newly available healthcare resource.

CONCLUSION

Our study of diverse patients’ access of clinic notes illuminates both the existence of disparities and grounds for optimism. While information technologies can perpetuate or even exacerbate the “digital divide,” they can also be potent arenas for healthcare institutions to champion equitable patient engagement. Achieving equity will require ongoing, targeted interventions tailored to the needs of diverse communities – including developing culturally sensitive communication strategies and creating environments where technology is a bridge rather than a barrier. While the Information Blocking Rule holds enormous promise for empowering patients through information, delivering on this promise means harnessing technology to build equity through every click of a digital portal.

Context Summary:

• Key objective: Are there disparities in patients’ access of clinical notes within oncology?

• Knowledge generated: While modest differences in access emerged around sex, age, and marital status, larger disparities appeared for ethnicity, race, and language: Black patients, Hispanic patients, and non-English preferred language speakers were, respectively, 37%, 15%, and 22% less likely to open at least one note, and opened 25%, 10%, and 18% percent fewer notes, compared to white, non- Hispanic, and English-preferred patients, respectively.

• Relevance (written by Dr. Madhusmita Behera): This study highlights the inequity in access to medical records in oncology by underrepresented patient populations. Addressing these gaps through interventions could improve patient engagement, quality of care and enable shared decision making.